Patent Application
20060274828
The invention relates to the field of audio/visual surveillance, and more particularly, but not byway of limitation, to such a system that is compact enough to be carried in a vehicle, such as a patrol car, and is capable of writing a high volume of data to digital tape such that high speed searching can be employed and is highly fault-tolerant.
Audio/visual surveillance systems that are sufficiently compact to be carried in a vehicle, such as a police or patrol car, are well known. These systems generally involve recording audio and visual information on a local recording system in the vehicle, transmitting the audio and visual information to a central command facility for review and/or recording, or combinations of the foregoing. See U.S. Pat. No. 6,037,977 issued May 14, 2000 to Roger Peterson. These systems also often include the acquiring and storing of location information, e.g., the geographical position of the patrol car. See U.S. Pat. No. 4,152,693 issued May 1, 1979 to Ashworth, Jr. These systems have been developed in response to the need for rapidly informing central command facilities, such as police headquarters, of emergency situations and the audio and visual details thereof, and the need for obtaining and preserving audio and visual evidence of crimes, emergencies, and other events that involve police action or participation. For example, to successfully prosecute an individual accused of a crime, the law of the United States of America requires that due process be shown. Audio and visual records can be of critical assistance in proving probable cause for stopping or arrest, and other due process elements.
Audio/video surveillance inherently involves a problem of data transmission and storage, because video data files are generally very large and surveillance must occur for significant periods of time, often days or weeks. Generally, this is addressed in surveillance systems by either saving only a few video frames per second, by storing frames for only a short time and then recycling the storage medium by recording over the previously stored data, or by storing or transmitting only portions of the surveillance data. See, for example, U.S. Pat. No. RE37,508 issued Jan. 15, 2002 to Taylor et al.; U.S. Pat. No. 6,211,907 issued Apr. 3, 2001 to Scaman et al.; and U.S. Pat. No. 6,456,321 issued Sep. 24, 2002 to Ito et al. A common solution to the capacity problem is to put the control of the recording devices at the fingertips of the police officers and/or headquarters and have them record only when it is required. See U.S. Pat. No. 6,037,977 referenced above. Surveillance systems also inherently require a system for rapid retrieval of data; and for this reason, most state-of-the-art systems data is stored on hard drives or other systems permitting random access. See, for example, U.S. Pat. No. 5,689,442 issued Nov. 18, 1997 to Swanson et al. However, hard drives are fragile if handled improperly, and downloading them without removing them takes so much time that it is unlikely to be done.
Audio/visual surveillance systems are employed in tens of thousands of patrol cars today. State-of-the-art systems, such as the device disclosed in the U.S. Pat. No. 6,037,977 patent mentioned above, give the police officer great flexibility with the multiple cameras and audio sources at his or her disposal. They include the latest technologies, including wireless transmitters, miniature cameras, removable hard drives, and geographical locators. Yet, the goal of having prompt communications with the officers in emergencies and reliable audio and visual evidence for use in court remains elusive. Often, in emergencies, police officers are responding to the situation and do not have time to activate the recording equipment. In most instances, due process evidence is not available because, by the time the systems are turned on, the probable cause evidence has come and gone. Even when the systems have been turned on, the resolution is often so poor that it either is useless or it takes a large amount of computer processing to enhance it to make it useable, or the hazards of police work combined with the fragility of high tech systems causes data to be lost.
In mission critical environments, such as those contemplated by mobile surveillance systems, tape is not a first choice, since, for all practical real-time purposes, the tape has been incapable of writing in a random access manner, unlike a hard disk that is a completely random access process. Typically, conventional streaming devices are problematic because losing any information for any reason at any point renders the remaining information beyond that point useless. For example, conventional analog or digital tape has stored thereon a directory or index of content stored on the tape, including start and stop information of content stored on the tape (e.g., streaming video). In the event that the directory or index information is corrupted or some portion of the content is destroyed, all content on the tape is lost. In the event that some portion of the content is destroyed, all content after the destroyed portion of the content is lost. In either situation, the lost content is generally unrecoverable. For these and other apparent reasons as understood in the art, tape systems have generally been avoided for use in mission critical environments, especially those utilized in harsh environments, such as mobile surveillance systems.
Conventional storage systems utilize storage mediums that are problematic for practical surveillance applications due to capacity limitations. As shown in Table A below, standard random access devices have limited capacity and/or have other serious limitations for practical surveillance applications used in harsh environments. DVD and CD-ROM have limitations in that recording is a once-only operation, and is not capable of start-stop recording. A hard disk can handle moderate shocks, but will be destroyed in a removable application if dropped. Although analog tape will continue recording during a shock, many undesirable artifacts are produced for several seconds after the initial shock.
To the extent that analog or even digital tape has been used for surveillance applications, conventional techniques for writing to these tapes are problematic for those interested in searching or seeking for content on the tapes. For example, it is generally understood that compression techniques may increase storage capacity of a storage media. In the event of using tape and writing recording time information in the compressed video content, a search of the tape for a particular time of the recorded video requires a system to uncompress the video, read the time stamp information, and determine whether the time stamp matches the time desired for the search. While such a search may operate up to four times normal playback time, in the case of having several hours of content stored on a tape, the search using the technique may take an excessive amount of time. Further, because compressed video using compression techniques such as MPEG-2 (Motion Picture Expert Group-2) is non-linear, searching using techniques other than conventional read search techniques results in an imprecise and timely manual search effort.
As described above, conventional techniques for reading compressed video include reading the video data and determining a time stamp value from the video compression scheme written thereon. This conventional technique for reading time stamp information from video compression introduces a few problems.
First, current tape deck technology offers the ability to read four times faster than real-time. While this enhanced reading speed offers improved searching capabilities, current tape decks are also capable of physically seeking at 400 times the speed of real-time. This means that reading the time stamps written to the digital tape using compression is a relatively slow process compared to the tape deck's ability to seek. Because of the non-linear writing using compression schemes, using the seek function of current tape decks on compressed video is simply not possible.
Second, the act of reading compressed video data to read time-stamp values assumes that both the recording and playback schemes have an intimate knowledge about how to parse and partially decompress the video to access these time-stamp values. While this knowledge between the recording and reading schemes appears to be straight forward, having to be limited to a tape deck having a certain format is problematic from a practical standpoint.
Third, a file directory is typically located at the front of a tape and includes a count value of tape marks that are used to indicate the start of files stored on the tape (e.g., data files). Continuous streaming of video onto a tape does not provide for such tape marks. The original intent of setting file marks, which are now called “set marks”, was to mark the start position of files so that a tape deck may quickly find a file on the digital tape by counting file marks. However, if the tape directory at the front of the digital tape is lost, the content of the tape is effectively lost because all context of what is on the tape is lost, and therefore fatal to further tape usage.
Accordingly, there is a need for a recording system that provides high resolution in a compact, rugged, and reliable system that stores high volumes of data in a high fault-tolerant manner that is capable of being searched at high rates of speed.
In overcoming the shortcomings of conventional storage systems for surveillance systems, the principles of the present invention provide for a reliable system that stores compressed video in a high-capacity, fault-tolerant manner that is capable of being searched at high rates of speed. The system includes markers that can be read independent of the compressed video, which markers are correlated to specific video recorded on the media. The markers can be read at a much higher rate of speed than the compressed video, thus allowing specific portions of the video to be found quickly.
In providing such a system, surveillance content maybe written to digital tape or other medium in partitions, preferably with directory redundancy and preferably with markers that maybe accessed independent of the tape content. The partitions form a function similar to the bulkheads in a ship; i.e., they limit the loss of data in case of corruption of a small part of the recording. The system also permits the streaming of multiple video signals, each from a different video source, onto a single digital medium, preferably a digital tape. A portion of each stream is written into each partition. The different streams may have different compression formats and different transfer rates. The recorded data is preferably self-authenticating. The surveillance system may be operated by accessing a web site and operating the system using a user interface on the web site.
The invention provides a surveillance system, comprising: a source of a video signal; a video signal compression system electrically connected to the source and providing a compressed video signal; a marker generator for generating markers independent of the compression, the markers indicative of specific content on the medium; and a digital video recorder electrically connected to the compression system for writing the compressed video signal to a recording medium and for writing the marker to the medium, the markers being readable independent of the compressed video signal. Preferably, the markers are timing markers recorded on the medium at predetermined time intervals. Preferably, the surveillance system also includes a marker read system for reading the markers. Preferably, the marker read system is selected from an electronic reader and an optical reader. Preferably, the marker read system generates a sound. Preferably, the marker read system comprises a timing marker counter for counting the timing markers without reading the compressed video signal. Preferably, the timing markers are spaced on the tape one second or less apart from each other. Preferably, the system includes a marker reader for generating sound signals from the markers. Preferably, the specific content comprises directory information regarding the location of data on the medium. Preferably, data comprises telemetry signals. Preferably, the medium is a digital tape. Preferably, the telemetry signals are recorded on the tape following the marker signals. Preferably, the recorder is a digital tape recorder and the recording medium is a digital tape having a semiconductor memory incorporated in it, the compressed video signal is written to the tape, and the markers are written to the semiconductor memory. Preferably, the surveillance system is mounted in a mobile vehicle. Preferably, the video compression comprises MPEG compression, which preferably is selected from the group consisting of MPEG-1, MPEG-2, MPEG-4 and H.264. Preferably, the video signals are high density (HD) video signals.
The invention also provides a surveillance method, comprising: generating a video signal containing surveillance images; electronically compressing the video signal into a compressed video signal; generating data associated with the compressed video signal; recording the compressed video signal and the data onto a digital tape cassette, the tape cassette having a semiconductor memory incorporated into it; and writing markers into the semiconductor memory, the markers designating where specific portions of the compressed video signal or specific portions of the is located on the tape. Preferably, the method further comprises reading the markers without reading the compressed video signal. Preferably, the generating data includes generating a start time and an end time associated with the compressed video signal. Preferably, the method further comprises: partitioning the compressed video signal into a plurality of partitions, each the partition including a portion of the compressed video signal; and using the markers to find a particular one of the partitions. Preferably, there are a plurality of the video signals, the electronically compressing comprises forming a plurality of streams of compressed video signals, each stream corresponding to a different source of the video signals, the method further comprising using the timing markers to locate one or more of the streams. Preferably, the data further comprises telemetry data associated with the video signal and the method further comprises using the markers to find the telemetry information on the tape. Preferably, the telemetry data includes time of day. Preferably, the generating a video signal is performed in a mobile vehicle. Preferably, the telemetry data includes one or more of the speed of the vehicle, the direction of the vehicle, the elevation of the vehicle, and an identification of the vehicle. Preferably, the video compression is MPEG compression, which preferably is selected from the group consisting of MPEG-1, MPEG-2, MPEG-4 and H.264. Preferably, the video signals are high density (HD) video signals.
The invention also provides a surveillance method, comprising: generating a video signal containing surveillance images; electronically compressing the video signal into a compressed video signal; recording the compressed video signal onto a digital tape; and writing timing markers, independent of the compressed video signal, onto the digital tape, the timing markers being spaced on the tape in a predetermined time pattern. Preferably, the method further comprises counting the markers written onto the tape without reading the at least one compressed video signal. Preferably, the writing timing markers comprises writing the markers in a periodic manner on the tape. Preferably, the timing markers are spaced two seconds or less apart on the tape and more preferably one second or less apart on the tape. Preferably, the method further comprises generating a sound from the timing markers. Preferably, the method comprises counting the timing markers without reading the compressed video signal. Preferably, the method comprises partitioning the compressed video signal into a plurality of partitions, each the partition including a portion of the compressed video signal; and using the timing markers to find a particular one of the partitions. Preferably, the method comprises receiving a time of day associated with the compressed video signal; determining the number of the markers from a position of the tape to the compressed video signal associated with the time of day; and moving the tape the determined number of markers and reading the compressed video signal. Preferably, the recording further comprises recording on the tape telemetry data associated with the video signals, and the method further comprises using the timing markers to find the telemetry data on the tape.
In another aspect, the invention provides a method of video surveillance, the method comprising: providing one or more video signals; compressing the one or more video signals to form a plurality of streams of compressed video data; and streaming a first of the video streams to via a first video channel while streaming a second of the video streams via a second video channel; wherein the first and second video channels each has a different transfer rate. Preferably, the method further comprises placing a time indication on each of the streams, which time indication is effective to permit the streams to be synchronized on playback. Preferably, the transfer rate of the first and second video streams differ by 10 megabytes per second (MBPS) or more. Preferably, the transfer rate is variable on at least one of the channels. Preferably, one of the video streams is a conventional density video stream and another is a high density (HD) video stream. Preferably, the compressing comprises comprising compressing a first of the video streams according to a first video compression standard and compressing a second of the video streams according to a second video compression standard, wherein the first and second video compression standards are different. Preferably, the first standard comprises MPEG-1 and the second standard is selected from MPEG-2, MPEG-4 and H.264.
In still another aspect, the invention provides a method of video surveillance comprising: generating a video signal containing surveillance images; generating self-authentication data; electronically compressing the video signal into a compressed video signal; recording the compressed video signal and the authentication data onto a digital medium; and self-authenticating the recording of the compressed video data using the self-authentication data. Preferably, the generating self-authentication data comprises generating a hash value. Preferably, the generating self-authentication data comprises generating time data from a GPS source or an atomic clock and the recording comprises recording the time data on the medium at intervals of one second or less. Preferably, the recording is performed at intervals of one-tenth of a second or less, and more preferably at intervals of one-one-hundredth of a second or less.
In yet a further aspect, the invention provides a method of operating a video surveillance system, the surveillance system including: a video camera providing a video signal; a video signal compression system electrically connected to the camera and providing a compressed video signal; and a digital video recorder electrically connected to the compression system for writing the compressed video signal to a recording medium; the method comprising: accessing a web site via a computer, and operating the surveillance system via a program located on the web site. Preferably, the operating comprises manipulating a user interface on the web site. Preferably, the user interface accesses only the predetermined local surveillance files. Preferably, the method further comprises customizing the functionality and look of the user interface. Preferably, the method further comprises providing built-in full SSL security Web server technology on the web site. Preferably, in the accessing is performed using a wireless system. Preferably, the video camera is located on a mobile vehicle. Preferably, the surveillance system is located on a mobile vehicle.
The above and other advantages of the present invention maybe better understood from a reading of the following description of the preferred exemplary embodiments of the invention taken in conjunction with the drawings.
Turning now to the details of the invention, patrol unit 102 includes a satellite signal receiver 108, a first audio source 110, a second audio source 112, a third audio source 114, a fourth audio source 116, a first video source 118, a second video source 120, a general input source 122, and an electronics box 130. Electronics box 130 includes a housing 134, a switch 138, which is optional and therefore is shown by dashed lines, an MPEG encoder 132, an MPEG encoder 136, and a computer 140. MPEG encoders 132 and 136 may be Mpeg-1, MPEG-2, MPEG-4, or H.264, and may have conventional resolution or high density (HD) resolution. High density resolution means any of the formats used or proposed for a resolution greater than the conventional NTSC resolution of 525 lines scanned at 29.97 frames per second with a horizontal resolution of 427 pixels. Computer 140 includes a solid state recorder/reader 127, solid state media 128, CD or DVD burner 129, parallel and serial ports 141, processor 142, RAM 143, a tape drive 144, a timing marker generator 145, a plurality of hard drives 146, a transmitter 147, and a receiver 148. Patrol unit 102 also includes antenna 150. Solid state recorder/reader is preferably a Flash or FeRAM recorder/reader, and solid state media 128 is preferably a Flash or FeRAM memory, though they may be any other suitable solid state system. Preferably, at least one of the media on which the video is recorded is removable; this maybe the tape 199, at least one of the hard drives 146A, or the solid state media 128. In some embodiments, there maybe more than one removable medium.
Command center unit 104 includes master sled bay 154, command center server 157, receiver 160, antenna 161, MPEG-1 monitor 166, computer 170, tape recorder 180, and DVDR recorder 182. As known in the art, master sled bay 154 is essentially a plurality of removable media drives, such as 151, 153, 155, and 156, along with control electronics. These drives may be tape drives, hard drives, solid state media drives, or any other drive for reading/recording on a removable media. Command server 157 includes processor 158, hard drives 159, RAM memory 162, MPEG decoders 163, and MPEG-1 decoder 165. Preferably, the hard drives 159 are organized into a RAID (Redundant Array of Inexpensive Disks) type storage system. Computer 170 includes monitor 172, electronics 174, including a processor and input and output cards as known in the art, and input device 176, which preferably is a keyboard. The various components of command unit 104 are connected by appropriate interfaces 190-194 as known in the art. Preferably, interfaces 190, 191, and 192 are SCSI interfaces.
In
The invention operates as follows. The satellite antenna 108 receives a GPS (Geographic Positioning Signal) and time signal T from satellites in orbit. How such signals are produced and received is well known in the electronics art. The GPS and time signals are fed to a serial port 141. The time signal is used to periodically set the clock of computer 140. Periodically, the GPS signal is processed, as known in the art, to produce geographic positioning information, which is buffered and recorded as will be described in detail below (
The encoded MPEG-1 signal is received via antenna 161 by receiver 160, processed by processor 158 as directed by software as described in more detail below, buffered in hard drives 159, decoded by MPEG-1 decoder 165, and displayed on MPEG-1 monitor 166. This process, as well as the activation of switch 138 in patrol unit 102, is controlled via computer 170. The MPEG-1 signal may also be stored via tape recorder 180 or DVDR/CDR recorder 182, or, as shown in
The removable media on which the MPEG signal is recorded is transferred to sled bay 154 by inserting it into one of removable drives 149 at the end of a patrol car shift or as required by operational policy. The data on the media is then processed by server 157. As discussed in more detail below, via a software program stored in memory 162, the instructions of which are processed by processor 158, the data is buffered in hard drives 159, depacketized, and decoded by MPEG decoders 163 into audio and video signals. The video signals are applied to monitor 172 to view the video while the audio signals are applied to speakers 178 and 179. Often, the decoded signals are also stored in some form. For example, utilizing computer 170, a user may select a certain portion of the recorded tape as being particularly relevant in a particular court matter. This portion may be depacketized and the MPEG data may be burned into a DVD disk via DVDR recorder 182. This disk may then be taken to court as evidence, without the need to have the entire command center 104 in court. The depacketized and decoded audio and video signals maybe stored by recording on tape via VHS recorder 180. However, since the VHS tape would not include authentication information (see below), such VHS tapes would generally be used for training purposes only.
In one embodiment, the tape drives 144, 151, 153, etc., are Sony AIT tapes, which are described in detail below, or may be an ADR™ tape drive manufactured by OnStream Data B.V., based in the U.S. and the Netherlands. These tapes utilize a completely enclosed cartridge. Several features of the preferred tape drive relevant to the invention are that the tape moves in a serpentine manner, the index is essentially in the middle of the tape, and the tape speed varies with the rate at which data is arriving. The index in the middle of the tape increases the speed at which the index can be written to and read. The variable tape speed allows the density of data on the tape to be maximized. For example, when the video is essentially static and little data is being generated, the tape slows down so that this little data is not spread over an unnecessarily large length of tape. This tape drive has rapid seek speeds, exceptional transfer rates, data reliability, and maximized media life. A single tape can store 60 gigabytes in the preferred mode, and up to 120 gigabytes if necessary. The ADR™ tape system has a 1019 bit error rate.
Turning to
As discussed above, the MPEG encoding and coding used in the invention are standard processes known in the art, and thus they will not be described in detail herein. A detailed description of the MPEG systems and processes is contained in “An introduction to MPEG video compression”, by John Wiseman; “Coding of Moving Pictures and Associated Audio for Digital Storage Media at up to about 1.5 Mbit/s”, ISO/IEC11172-2: Video (November 1991); and “Generic Coding of Moving Pictures and Associated Audio Information: Video”, ISO/IEC13818-2 Draft International Standard November 1994), all of which are hereby incorporated by reference to the same extent as though fully disclosed herein. However, the packetizing of the encoded MPEG data and the arrangement of the packets in the data stream provided by the invention are novel. An illustration of an audio/visual data stream 501 as it can appear on a tape 199 according to the invention is shown in
The audio/visual packets VA and VB preferably are of variable length, depending on the complexity of the information being captured. For example, if the scene being photographed is rapidly changing, the packets will be longer, and if the scene being photographed is static, the packets will be short. In the preferred embodiment of the invention, the longest packets are 32 kb-20 bytes and the shortest packets are 21 bytes. The packets are created and placed in the stream by a protocol that depends on the amount of data in a buffer and other efficiency factors. Those skilled in the art of communication buffers will be able to create appropriate packets; thus, the details of this protocol shall not be discussed herein. Many different such protocols may be used. In the portions of the data stream shown in
The inventive file system 610 and how it operates the hardware described above is illustrated in
In the file system 610 according to the invention, the data generated by application threads 712 and 132, which in the application herein are encoded MPEG audio/video channels, is directed to per file write cache buffers 720. Each application thread, that is, each MPEG channel, is directed to a different buffer. The V1 channel is directed to buffer 724 and the V2 channel is directed to buffer 726. In the embodiment shown, there can be up to four application threads, as four buffers are shown; however, more than four application threads and more than four per file write cache buffers may be used. The data in the buffers 720 is organized into VA, VB, and D packets and interleaved into a data stream 501 by write multiplexer thread 730. The data stream is directed to streaming write cache buffer 734. The purpose of streaming write cache buffer 734 is to eliminate any differences between the flow of the data stream and the operation of tape 144, which differences can arise in the mechanical operations of the tape drive. For example, the tape must pause in accepting data when it reaches the end of the tape and reverses. During this time, the streaming write cache buffer will collect and hold the streaming data. The write streamer thread 736 forms the final streamer thread and directs it to driver 636, which delivers it to tape 144. The data stream is parsed continually in write streamer thread and tracking data, such as the location of GOP (Group of Pictures) headers, and year/month/day information is stored in directory cache 738.
In another embodiment of the invention, the MPEG GOP headers are modified to add the geographic and year/month/day information. Other digital information also maybe added to the GOP headers. In this embodiment, the decode system software is modified to read this information. In another embodiment, this data is periodically added to the MPEG-1 stream with a marker to permit it to be easily found. This information is preferably displayed directly on the screen on the monitor with the video, although this feature can be turned off.
Content verification of the video, audio, and GPS data is done via computation of an MD5 Message Digits 5) hash on the data streams as they are output from the hardware encoding devices. To insure that encoded data is not modified and re-hashed, an administratively designated non-retrievable pass-code is assigned to each Mobile Unit before it enters the field. The resultant hash codes, a combination of data and pass-code, is stored with the directory data and can be used to tell if any of the data streams have been modified. MD5 hash codes (128-bits) are computed over video GOP (Group of Pictures) intervals; i.e., they are constructed from all video, audio, and PS encapsulation data between GOP headers. This process is not an encryption or watermarking scheme. The message digest function is also sometimes referred to as a one-way hash function.
The MD5 hash function is a one-way algorithmic operation that transforms a string of data of any length into a shorter fixed-length value, usually 128 bits or 16 bytes long. The algorithm is coded in such a way that there is a negligible probability that any two strings of data will produce the same hash value. If just a single piece of data is changed, a different hash value results. At anytime, data integrity can be checked by running a utility verification program supplying the original pass-code, which program is generally referred to as a checksum procedure. That is, the data integrity can be verified by running a hash operation on the data and the private pass-code, i.e., the one assigned to the patrol car when it enters the field. The resultant hash value is compared to the hash value stored in the data. If the two values match, that data has not been altered, tampered with, or modified in any way, and the integrity of the data can be trusted. This comports with the “best evidence rule” and authentication requirements used by courts. The MD5 algorithm is a well-known standardized algorithm Thus, it will not be further discussed herein. It is generally believed that it is computationally infeasible to duplicate an MD5 message, or to produce any pre-specified MD5 message.
Just before the tape is removed, the information from directory cache 738 is written to the QFA portion of tape 199.
The operation of file system 610 in command center unit 104 is the reverse of its operation in the patrol car. The data stream is read out of the tape drive via the device class drivers 636 and delivered to the read streamer thread 744. The read streamer thread 744 directs it to streaming read cache buffer 748, which smoothes out any discrepancies between the flow of data and the mechanical operation of the tape drive 144. The data stream is demultiplexed by read demultiplexer thread 750, and the data associated with each application thread is cached in the appropriate one of per file read cache buffers 760. Namely, the data from the V1 MPEG channel is cached in buffer 762 and the data from the V2 channel is cached in buffer 764. The buffers then stream the data to the corresponding application thread, which in the exemplary embodiment is the corresponding decoder 163 or 710. Those skilled in the art will recognize that the hard disks 146, together with the microprocessors 142, under software control, act as the buffers and multiplexers of the patrol car side, while the hard disks 159 and microprocessor 158, under software control, serve as the buffers and demultiplexer of the command center side. These buffering and multiplexing functions are well understood in the computer art and, therefore, will not be described in detail herein. The file system 610 also includes statistics and interval time thread 766. Thread 766 provides a set of private I/O Control (IOCTL) codes that allow an application program to set options and gather statistics on file system performance. The statistics gathered can be used to tune cache buffer sizes and optimize aspects of the read/write streaming algorithms.
The streaming write capabilities of sequential access devices dictate that write operations always be performed at the current End-of-Data location. This knowledge forms the basis of the above-described two-stage write cache architecture. Write caching is done on a per-file basis to decouple slow sequential device access times from the application thread requesting the synchronous write operation. For synchronous writes, the application thread is blocked only until write data has been queued to the write cache buffers. The write queue is serviced by an internal worker thread that is directed by a multiplexing algorithm, which places its results into the device's multiplexed-write queue. The multiplexed-write queue is serviced by an internal worker thread that is directed by a streaming algorithm optimized for device write streaming. To mitigate paging area contention, internal cache areas are backed by temporary files on disk-based file systems, preferably on non-paging NTFS drives. Data from multiple file write sessions is multiplexed at the media block level such that the average data rate for a given file is maintained over time.
The streaming read capabilities of sequential access devices permit random read access. Because of this capability, and the need to permit multiple simultaneous reads, the read cache process is not an exact mirror of the write cache process. Read data is read-ahead streamed off the device and placed into the multiplexed-read queue. When a specific file is opened, its data is broken out from the multiplexed-read queue into its own read cache buffers. The read cache buffers and the multiplexed- read queue are filled by a special algorithm optimized to give increased read performance priority to files that were opened first.
The system 100 uses a TCP mediated distributed architecture providing flexible scalability through the addition of modular components. This network-based approach uses TCP/IP point-to-point connections for commands that don't require synchronization (i.e., configuration and monitoring). For synchronized activities, a UDP connectionless protocol is used to broadcast commands providing more accurately synchronized record/play/stop/pause functionality across the distributed architecture.
A feature of the invention is that sequential write performance is equivalent to that of writing to a physical disk drive. Read performance is based on a number of factors. If the files being read were written at the same time, i.e., their blocks were multiplexed close together, then read performance is equivalent to that of reading from a physical disk drive if the average aggregate read data rate is not greater that of the underlying sequential access device. If aggressive head movement or volume exchange is required to obtain their data blocks, then read threads are delayed until such data can be located.
Other details of the system 100 are as follows. Since the geographic information is available via the MPEG-1 stream as discussed above, this information can be used to locate the patrol car in an emergency. It is also evident that the invention can be used to construct all-points news bulletins, notification to other departments, and has many applications for training purposes. The system 100 includes electronic circuitry, software, and processes to provide remote power-on/boot and power-off from the patrol car dashboard, full cycle boot at power up without user intervention, vehicle ignition-controlled start and shutdown, programmable shutdown, battery fed continuous operation after ignition shut down, vehicle speed, direction, and location integration via the GPS information, out-of area and failure notification, and status of storage available indication. System status lights and failure lights are dashboard mounted. The central command center features include: on-demand wireless communications with mobile units for audio and video “real-time” viewing, post-event playback and review capabilities, multiple unit synchronization, full VTR controls with added search capabilities, and post-production capabilities. Existing MDT, CDMA, or cellular technologies can be incorporated for the wireless transport of the MPEG-1 signal, geographic and time data, patrol unit and shift, and other significant information. Built-in diagnostics monitor video encoder status, battery condition, power supply output, system operating temperature, and many more system conditions. The command server 157 has been designed using a single board computer (SBC) with the Coppermine™ 700 MHz Pentium III Processor, and up to 512 MB of Random Access Memory (RAM). The SBC is designed to be a functioning “mini” computer built onto a PCI card. In the event of a failure of the board, CPU, or RAM, the card is simply replaced without dismantling the entire system. As indicated above, the command server can be equipped with CD-ROM recorders, DVD recorders, or digital tape recorders for long-term archival depending on client needs. The Raid system 159 is preferably a RAID 01 system with mirrored hard drives. The cameras are high-resolution color for normal light IR with monochrome imaging for low or no light situations. They both have wide-angle camera lenses and include a composite video splitter, i.e., two inputs to one composite output. The front facing color camera 118 features a ½-inch CCD capable of capturing an NTSC image with 480 lines of horizontal resolution. The minimum illumination is 1.0 Lux through an auto iris F/1.2 lens. The rear facing wide-angle color camera 120 features a ¼-inch CCD capable of capturing an NTSC image with 350 lines of horizontal resolution. The minimum illumination is 2.0 Lux through an F/2.0 lens. The interior microphone 114 is sensitive to 1V/Pa @ 1 kHz (−2.5 dBV +/−4 dBV) and has an output impedance of less than 150 Ohms. The voice input distance ranges from 7 cm to 1.5 m to accurately capture all audio within the seating area of the patrol car.
The system includes automatic file naming with unit number, date, time, and shift, which is included in the QFA section.
A feature of the invention is that the streaming tape recorder capable of a data rate equal to or greater than the aggregate recording rate permits VCR-like functionality in a digital tape recorder of much higher resolution.
Current mobile surveillance systems record to analog VHS tapes or camcorders. The problem with analog VHS tapes is that the video quality is poor and most tapes record for only a couple of hours. Some manufacturers claim much longer recording times of up to eight hours, but those are typically at very slow frame rates of recording, making for jerky movements and poor image quality. A feature of the invention is that DVD-quality video results. Additionally, digital tapes can be reused for 30,000 cycles and the shelf life for digital tapes with no degradation in quality approaches thirty years as compared to 30 cycles and one to five years for analog tapes.
It is a feature of the invention that the data is streamed to digital tape in real time. Except in cases where the tape is changing direction or some similar event, the data is processed immediately and passed to the tape, rather than being stored for a significant time, for example, for a time greater than normal computer processing time, and then processed later. Real time also means that, from the perspective of a human being, the transfer to tape usually would appear to be instantaneous. A related feature is that, in the system of the invention, the digital recorder and digital tape comprise the primary storage system rather than a backup storage system.
It is another feature of the invention that the system 100 captures full-motion video. Full-motion video is any video that captures at least 24 frames per second and more preferably at least 29 frames per second. As known in the art, the full-motion video NTSC standard is 29.97 frames per second. A related feature of the invention is that the system 100 at the same time captures full-frame video, which means any resolution of at least 720×480 pixels. A further related feature of the invention is that the system 100 can capture at least eight hours of full-motion, full-frame video on a single digital tape. A further related feature of the invention is that the system 100 can capture at least eight hours of two full-motion, full-frame videos on a single digital tape.
A further feature of the invention is that each data packet, such as 504, is independent. By “independent”, it is meant that at least a portion of the audio and at least a portion of a video frame can be reconstructed from a single packet. In the prior art, if a system were downloading a video file and the process was interrupted before it was completed, the result would be unintelligible. However, in the system 100 according to the invention, the packetization system and process results in a single packet being intelligible. Of course, the more packets that are received, the more of the sound and video can be constructed. Thus, if only a portion of a tape is available, say due to a fire or other catastrophe, useful information can still be obtained from the tape.
Another feature of the invention is that the mobile system 102 is designed for use in a dynamically changing environment. The basic unit operates in temperatures in excess of −25° C. to 81.1° C. with an optional electric temperature controlled environment. Depending on configuration, operating temperatures required for the power supply are −25° C. to 81.1° C. or −40° C. to 80° C.
The above describes a novel vehicular surveillance system that permits a full shift of two MPEG channels of full-frame, full-motion audio/video to be captured on a single cartridge tape. The system for the first time permits 24/7 patrol car surveillance at high resolution. Now that the system has been created and disclosed for use in patrol cars, it is evident it will have applications in many situations in which a compact, high-resolution surveillance system is desirable. For example, it will have applications in airplanes, trains, ships, and other vehicles. Thus, wherever the terms “patrol car” or “police car” have been used above, any other vehicle may be substituted. It also will find use in many security applications. It is believed that the invention will make digital tape cartridges a preferred primary storage device. Examples of such applications are as follows. The system of the invention could be used in a manufacturing facility, such as an automotive assembly line or an integrated circuit manufacturing facility for quality control purposes. In such manufacturing processes, defects often occur that are difficult to find the reason for. Since it usually is known when the particular vehicle or part was manufactured, a library of 24/7/365 tapes would be useful in tracing and correcting defective processes or systems. Another example is any test operation, such as the test of a jet fighter or the destructive test of a system. Since it is often not known when the object being tested will deviate from specification, a 24/7/365 surveillance system would be useful. The system can also be useful in an operating room to record an operation from many different angles for instruction or legal purposes. It may also be used in stores, government and public buildings, and anywhere else that surveillance systems are in use today.
The surveillance system 100 according to the invention was developed to provide an improved patrol car surveillance system. To achieve this goal, many novel components had to be developed. Now that the system has been built, it is evident that many of these elements will have important uses in other applications. For example, the file system 610 according to the invention that streams data to digital tape will be useful in many instances in which rapid streaming of sequential time and/or geographic synchronized data is desired. For example, it is useful in database logging, ISP logging, transaction logging, firewall logging, backups, general audio/video encoding, and data acquisition.
A feature of the file system 610 is its ability to multiplex data from several streams into one bundled stream that is then stored on and retrievable from the tape drive. Another feature of the file systems 610 is the ability to access the tape drive from a PC as a local drive letter or as a Universal Naming Convention (UNC) mapping across a network. In addition, the relative cost of tape drives and their media is less, on a per gigabyte basis, than the cost of hard drives. These features move the tape drive from its traditional position as a data backup product to that of a primary storage medium for many applications. The types of applications that are particularly targeted are: (i) those in which the data does not need to be accessed often; (ii) those in which data does not need to be written onto the tape and accessed at the same time; and (iii) any of the foregoing applications that would benefit from a removable medium.
The principles of the present invention represent a paradigm shift with respect to patrol car surveillance systems. The prior art patrol car surveillance systems were seen as tools to be subjectively used by police officers. The principles of the present invention view surveillance systems as being objective tools of administrators, prosecutors, and courts.
In addition, the principles of the present invention advance the art by overcoming conventional surveillance system problems by recognizing that the way to avoid having evidentiary gaps in the audio/visual record is to have high resolution audio/visual recording operating at all times that a police car is on patrol, 24 hours a day, 7 days a week, 365 days a year. With the prior art video systems, this would immediately lead to data overload. However, as described herein, the above requirement does not mean that the audio/visual system has to be able to store scores of hours or days of data in the vehicle, because patrol officers always work in shifts that generally are of from 8 to 12 hours in length. If changing the data medium is made simple enough, it can become a routine part of the shift change, and operate repeatedly and reliably.
While the replaceable hard drives that have become part of most surveillance systems today are advertised as being simple to use, in fact, few people can routinely and repeatedly perform the change and/or perform a downloading operation without incident. Further, the fragility of hard drives and the hazards of police work make the use of such drives problematic in the patrol car environment. Changing a tape cartridge is something that most people today can do repeatedly and reliably. Further, tape cartridges are rugged and tape data is rarely inadvertently destroyed. In accordance with the principles of the present invention, an audio/visual surveillance system records data and/or content to a tape cartridge within the vehicle. In one embodiment, the tape is digital tape. This provides an essentially fail-safe system in which data is reliably and routinely transferred to a central storage system at the end of each shift. The system includes a cartridge tape storage sled bay at the police headquarters or other facility to which officers return at the end of a shift. At the beginning of the shift, each officer is provided a tape cartridge, which they insert in the recorder in their patrol car. At the end of each shift, the officer simply removes the tape cartridge from the patrol car and inserts it in the tape storage sled. The rest is automatic.
The MPEG-2 video/audio compression standard is well known in the movie and video art, though it is usually associated with DVD systems. The MPEG-2 standard provides the high-resolution, dense storage associated with home DVD systems. The system and process permits the direct recording of MPEG-2 audio/visual data to a cartridge tape in a patrol car. Further, with the development of MPEG-4, this and other compression techniques may alternatively be utilized for surveillance systems in accordance with the principles of the present invention. In this disclosure, any reference to MPEG-4 includes H264, MPEG-4/H.264, MPEG-4 Part 10, H.264/AVC, or any other designation that is associated with this standard, as well as any other part of MPEG-4.
The system also provides for wireless transmission of audio/video directly from the patrol car to the central command center or headquarters. Since wireless transmission does not presently have a broad enough bandwidth to support real time streaming of MPEG-2 audio/visual, the system also provides for MPEG-1 encoding of an audio/visual signal, which MPEG-1 encoded signal is buffered, preferably in a RAM or hard drive, and then may be transmitted on command. Preferably, the MPEG-1 encoding and wireless transmission can be initiated from either the patrol car or from the central command center via a wireless link
The system also provides an arrangement of audio and video sources that is designed to capture most, if not all, events of interest. There may be two or more video sources, one of which captures events outside the patrol car and the other of which captures events inside the patrol car. There may be three or more audio sources, one of which captures audio inside the patrol car, another which is on one officer's person, and a third that captures the radio exchange with the dispatcher. If two officers are present, a fourth audio source may be on the second officer's person. One video signal and two of the audio signals are encoded in a first MPEG-2 channel, and the second video signal and the third and fourth audio signals are encoded on a second MPEG-2 channel.
The two MPEG-2 signals are buffered, formed into data packets, and multiplexed into a single data stream. The multiplexed data stream is preferably buffered to remove asynchronies between the tape movement and the incoming stream, and then is recorded on the tape.
MPEG-4, MPEG-2, and MPEG-1 contain time synchronization data. As known in the MPEG art, each frame contains synchronization information. Further, the synchronization data is keyed to a GOP (Group of Pictures) header that occurs regularly, for example every 15th frame, in the MPEG data, or approximately every one-half second. This synchronization data time correlates the individual MPEG frames. Geographic location data and, preferably, absolute time data may be acquired via a satellite link or otherwise. Hour/minute/second data are automatically incorporated into the MPEG data as known in the MPEG art. The tape maybe parsed and the location of each GOP header is found. This GOP header location information and year/month/day data are cached in a buffer and recorded in a tracking location on the tape. Using the year/month/day data and the MPEG synchronization data, each frame can be accurately time referenced. The absolute time signal is used to periodically update the clock of the system computer. In this manner, each frame can be time referenced within a fraction of a second. The geographic data maybe recorded in a special digital frame that is recorded regularly on the tape, preferably every five seconds. This digital frame may also include information such as if and when the patrol car shotgun is removed from its rack, radar data, and any other special data that a user may desire. All of this data may also be recorded in a header to each data packet so that, in case of system failure, all the geographic and time data can be reconstructed.
Once the tape cartridge is inserted into the sled bay in the central command center or other location, the system, software, and method of the invention permit the audio/visual data to be easily retrieved, monitored, synchronized with other data, stored, and archived. This is facilitated by the fact that it is encoded via the MPEG-4 or MPEG-2 standard. The data on the tape may be transferred to a hard drive of a command server with a form of RAID data storage. If the data is to be monitored, multiple videos can be synchronized and viewed at the same time. In one embodiment, up to four videos can be viewed at the same time. For example, if four police units were at an event and recorded the event, the event can be viewed from four different angles. The data can also be decoded and transferred to any desired medium, for example, an analog tape or a DVD disk.
The system permits the tape hard drive cache system to be accessed as a universal naming convention (UNC) drive, which is most commonly implemented as a letter. That is, using conventional software programs, such as Windows™, the invention permits the tape hard drive cache system to be designated as the “D” drive, for example.
The MPEG-1 low-resolution data stream is also buffered in the central location on a hard drive of a server. It maybe decoded and monitored directly, or it maybe decoded and stored on any suitable medium, such as a VHS recorder. Via the tracking data, it may be synchronized with MPEG-1 data from other units, or at a later time, with MPEG-2 data in storage.
An authentication process that ensures that the recorded audio/visual evidence will be acceptable to the courts may also be utilized. In one embodiment, a private pass-code is assigned to each patrol car as it goes in the field. This pass-code is used to generate a verification code that is stored on the tape. This verification code can be used to authenticate the data at any time by running a verification procedure, preferably a checksum procedure.
Essentially, all audio/visual information associated with a patrol car maybe reliably captured, monitored, correlated, stored, retrieved, and authenticated in accordance with the principles of the present invention. For example, a common occurrence today is that a suspect or criminal will claim officer brutality and point to bruises as evidence of the charge. Often, however, the bruises are self-inflicted after the person has been confined within the back seat of the patrol car. Each such charge, even if false, usually costs the jurisdiction a significant amount of money, on the order of $25,000.00, in investigating the charge and prosecuting it, if necessary. The invention will go a long way toward reducing and/or eliminating such expenses.
One or more video recorders 808a-808n (collectively 808), embodiments of the video sources 118 and 120, operate as surveillance devices. The video recorders 808 and GPS and general information devices 809a, 809b, and 809c (collectively, 909) maybe wired and connected via a physical cable, such as 810, or wireless and communicated across a wireless link 812. A video server 814a may receive video signals 816 and 818, for example, from video cameras 808a and 808b, respectively, and other signals 817 representative of general data, such as event information, as well as GPS data, to be compressed into a compressed video signal 820 (e.g., MPEG-4 video signal) and stored on a digital removable media, such as a tape drive 822, a semiconductor memory drive 823, or a removable hard drive, in accordance with the principles of the present invention. In addition, the compressed video signal 820 is preferably stored on a hard drive 815a in server 814a or other medium to either as a backup or as a primary storage device. A timing marker generator 821a, 821b is also included in servers 814a and 814b. The video recorders 808 may output the video signals as digital signals in a digital stream, packet format, or otherwise, or as an analog signal to be converted into a digital signal at the video server 814a or other controller (e.g., electronic box 130 of
A hub 830 may be integrated into the network 800 and be configured to enable users on the network to access content stored and maintained by the video servers 814a and 814b (collectively 814) accessible to the hub 830. As shown, there maybe a number of remotely located computers 832 configured to engage the network 800. In one embodiment, the network 800 is the Internet. The computers 832 may access anyone of the video servers 814 configured on the network as understood in the art. Accordingly, people operating the computers 832 may access content (e.g., surveillance content) that is stored on digital tapes or other media for review thereof in accordance with the principles of the present invention. Alternatively, a camera 876 and other surveillance devices may be connected to a computer or workstation 878 to provide a surveillance subsystem 874.
A multiplexer 910 is configured to receive the video signal 906 and audio signal 908 from each of the splitters 904 and form a multi-channel content stream 912 that includes the video signal 906 and audio signal 908. The multi-channel content stream 912 is input into a sliding window 914 for use in writing onto a medium, such as digital tape, a removable hard drive, or other media. The sliding window 914 may be a processor executing software configured to operate as a sliding window as understood in the art.
The input crossbar 1004 outputs a digital signal 1005 including YUV2 frames and audio signal to a digital video/audio compressor 1020. The digital video/audio compressor 1020 receives the digital frame and applies a compression scheme to reduce the data to a manageable size. The compression scheme may be any compression scheme utilized to compress digital video signals as understood in the art. The compressed video signal is output from the digital video/audio compressor to an optional multiplexer 910. The multiplexer 910 is configured to receive multiple video streams and combine them into a synchronized or multi-channel content stream 912. The multi-channel content stream 912 is buffered to a hard disk using a sliding window 914, where video segments are added to the back of previously stored video segments. The multi-channel content stream 912 is separated into disk segments at 1028. At 1030, the video stored onto the hard disk is read from the front. A sliding window segment reconstructor 1032 reconstructs the video and generates a video stream 1034, which is communicated to a tape system 1036. The tape system 1036 writes the incoming video stream 1034 to digital tape cassette 840 on the tape drive 822 (
Video/audio module 1102 includes encoders 1104, control electronics 1106, and abstract encoder module 1108. Abstract encoder module 1108 is designed to be compatible with all or nearly all off-the-shelve video encoders. Thus, many different video encoders, such as a direct show encoder, a canopus encoder, a DVD plus encoder, a Vweb encoder, a solid state encoder, or anyone of future encoders that become available maybe used with the system 1100. The customer can specify which encoder is preferred, and one or more of the encoders shown may be incorporated into a specific system. Control module 1106 permits the compression type, such as MPEG-1 through 4, to be set, either variable or constant bit rate to be set, and the specific bit rate, such as 750 K per second through 25 MEGS per second, to be set. Other video encoder parameters may also be set as known in the art. Input module 1102 also includes user activated inputs 1103, such as initialize, de-initialize, start and stop. The encoded signal is output to buffer 1114 at output 1110. Data input into buffer 1114 circulates in the buffer, is queued, and is output at 1120 as required to create the organized partitions described below.
At the same time as the video is being compressed and organized, the GPS and general data is being collected and organized in general data module 1176 and GPS data module 1130, respectively. General data module 1176 includes a digital input/output module 1180, a mission critical unit (MCU) 1178, and a serial input/output module 1177. Digital input/output module 1180 provides vehicle speed, vehicle direction, vehicle elevation, vehicle or camera identification, and other information as specified by the customer. MCU 1178 is also customer specific. It includes a one of more of a variety of relays and switches which develop a specified voltage in response to a specific triggering event, such as shotgun removed from cradle, acceleration of a vehicle that is indicative of a chase, movement of a vehicle as may be indicative of an explosion or an accident, and manual switches such as to indicate a traffic stop, a robbery in progress, or other mission specific data desired by the customer. Data from DIO module 1180 and the combination of serial input/output module and MCU module can include any other information useful for surveillance, law enforcement, emergency response, other application-specific information and other information associated with the video/audio being recorded. GPS data module 1130 includes a serial input/output unit 1131 and GPS data unit 1132. GPS latitude and longitude telemetry information is output at 1133.
Merged event/MPEG writer 1126 receives input from outputs 1185 and 1133, merges it with video data output at 1120, and feeds it to merged video/audio/GPS/ general data buffer 1136 under control of inputs 1124 which include Find Next Group of Pictures (GOP), Write New MPEG file, and Event/Telemetry Data In, the latter which is a signal indicating that non-video data, such as GPS data, event data or other general data associated with the particular GOP and MPEG file is available to place in buffer 1136. Buffer 1136 is preferably a hard disk or semiconductor storage, but it also may be any other suitable media. In buffer 1136, separate streams 1140, 1141 through 1142 are set up, with each stream corresponding to a particular camera 808a etc. or other video input device. Each stream includes an MPEG header, such as 1139, and MPEG queue files, such as 1138, are shown in
The signals from storage media 1170 is fed through output 1162 to NMEG In/Preview input 1193 to output system 1186. Output system 1186 includes a decoder module 1187 including video/audio decoders 1188, on-screen display control electronics 1194, and abstract decoder module 1190. Abstract decoder module 1190 is designed to be compatible with all or nearly all off-the-shelve video decoders. Thus, many different video decoders, such as a direct show decoder, a canopus decoder, a DVD plus decoder, a Vweb decoder, a solid state decoder or anyone of future decoders that become available may be used with the system 1186. The customer can specify which decoder is preferred, and one or more of the decoders shown may be incorporated into a specific system. On-screen display control 1194 includes the inputs 1195 to control the on-screen display, which inputs include initialize, de-initialize, bit map display, text display, and flip page. Input module 1102 also includes user activated inputs 1192, such as initialize, de-initalize, step n, start and stop, pause, and set position. Other video encoder parameters may also be set as known in the art. The encoded signal is output at output 1191 to video display module 1196, which generally is a computer, and thus the video may be either a Windows™ display 1197 of a Linux video display 1198.
Control module 1199 controls the digital settings for the system, preferably in XML or INI, and feeds control signals to the rest of the system via outputs 1199a, 1199b, 1199c and 1199d.
Partition information 1204a preferably includes a stream map 1208a, telemetry 1210a, roster information 1212a, and tape position information 1214a. It should be understood that the partition information 1204a may include different and/or additional information associated with the digital video stored in a particular partition. Each or the n partitions will include this information.
Stream map 1208a preferably includes video stream information 1216a-1216m (collectively 1216) that includes information associated with the digital video stream. The stream 1208a preferably further includes start time 1218a and end time 1218b of the video segment. Different and/or additional information associated with the digital video stream maybe included in the stream map 1208a.
The telemetry 1210a preferably includes data indicative of physical or other parameters during the recording of the surveillance video. In one embodiment, the telemetry 1210a includes an event list 1220 (e.g., shotgun removed from cradle, chase, robbery in progress, accident), GPS location in latitude/longitude 1222, speed 1224, direction 1226, elevation 1227, time 1228, date 1229, and camera or other video input or vehicle ID 1230. Other parameters may also be recorded, including temperature, lighting conditions, vehicle number, or any other parameter useful to providing information associated with the surveillance video at a later time.
Roster 1212a preferably includes time parameter 1232 and comments 1234. Comments 1234 may include comments entered by an operator on a computer associated with a video source, for example. Time parameter 1232 is preferably the time the comments were made, or other time associated with the comments. Different and/or additional information may be included in the roster 1212. The record is searchable by any information in the partition directory, including, but not limited to, any information in the stream map, the telemetry, and roster.
A feature of the invention is self-authentication. By self-authentication is meant that the recording can be authenticated, that is, shown to have not been tampered with or forged, with only the record itself and the playback system. That is, only the recorded tape, hard drive, solid state memory, or other medium on which the video is recorded and the playback system with authentication software are required to authenticate the record. For example, U.S. Patent Publication No. 2002/0131768 published Sep. 19, 2002, discloses an authentication method that uses encryption and requires a court or other authenticator to have an encryption key to authenticate the record. Thus, that system is not self-authenticated since it requires something outside the record itself and the playback system for authentication. One example of a self-authentication method is the hash value 534 discussed in connection with
To enable fast seeking of video without having to read and uncompress the compressed video and stream map 1208a to read the video segment time 1218a-1218b, one embodiment includes markers 1238a-1238r (collectively 1238). In the embodiment shown, the markers are sound or optical markers placed onto the digital tape with a regular period. A one second period is shown, but other periods may be used, such as every half-second or every 1.5 seconds. In this embodiment, these markers 1238 are real-time markers indicative of the relative time after recording of the surveillance video starts and are independent of the digital video signals. In other embodiments, the markers may be markers that are generated by an algorithm, or markers pointing to the location of the directory information and stored in memory 1604 (
The system, in accordance to the principles of the present invention, incorporates another level of fault tolerance, which is accomplished with logically breaking a tape into multiple independent partitions. Typically, the tape is segmented into ten or more independent tape partitions. By segmenting the tape, the tape system can withstand even a massive failure on tape (such as a wrinkled tape). A conventional tape system cannot continue with such a failure. However, with a multi-partitioned tape, the system deems this partition as unusable and simply skips to the next partition. During recording, there is no loss, as it automatically skips. During playback, the maximum possible loss with such an error would be a partition worth of blank video. In most cases, data written up to the error situation is intact. Since the tape system skips immediately upon trouble, the data resumes on the next partition.
Playback and Display
The system according to the invention can track a single incident from multiple camera angles and, preferably, with multiple audio tracks, with the time and location synchronized. The record is searchable by time, date, vehicle ID, GPS location, and event. Any of these elements, such as GPS location, can be displayed for each stream, i.e., each camera. Zoom and pan can be controlled individually for each stream during playback. The system also permits frame by frame search, generally using time as a locator. Brightness, contrast and saturation can also be controlled individually for each stream, Likewise, each audio channel can be individually controlled. Segments of a recording can be easily clipped and copied to a disk or other medium.
Interruptible Video Stream
As many users have experienced, it is also usually fatal to have any kind of data corruption in digital data. Any corruption to the data stream itself normally would prove to end a fate similar to any program that has a few bytes in the wrong place. Since the present system is intended for streaming video, advantage is taken of the streaming MPG standards, which periodically insert start codes into the video stream. These are markers into the video data indicating the re-sync point. These start codes find and re-synchronize the video stream should a portion of the data be corrupt. The longest video length that could be lost after a corruption is a maximum ½ second, as the system automatically picks up where it left off once the corruption area has been passed.
With the redundant nature of the tape index and the ability for the stream to re-synchronize again even after a major tape corruption, the present system is extremely fault tolerant and suitable for any mission critical application.
Full Index Reconstruction
Should the need arise to reconstruct a new index, the tape system goes into full fast forward mode looking only for tape markers. Once a tape marker is found, the tape automatically slows down and starts reading the first block after this tape marker. This block is the redundant demarcation block, which contains all the information needed to retrieve the stream's name and information. Should this first block be corrupt, the tape simply continues to read until it reaches another demarcation block, which is normally just a few seconds ahead, until it reaches a good block. The index in this system does not record the relative byte offset to the stream, but counts the tape markers to reach it. Since the tape can seek at its highest speed, and because it does not need actually to read the tape to search for tape markers, it does not get hung up on the first tape read error. Also, the tape directory cannot be corrupt, bad, or missing due to the fact that there are streams on the tape, which are enough to fully re-construct an index; the demarcation blocks hold a directory entry for the stream.
This process is done until the end of tape is reached (indicated by a special tape marker). At this point, the index reconstructed in memory is compressed and written to the EEPROM on the tape once again.
Tape Marker Seeking
Once an entire tape index is created, it is known how many tape markers to count either forwards or backwards from any point on the tape to quickly seek to the stream being sought, which is the same as performing a byte count as understood in the art. There is some optimization information also stored with the index, gleaned from the next and previous demarcation blocks. This allows for predetermining a particular stream and to seek within that stream.
Even without a tape index ever being created, striping a tape to hard disk can happen instantly if starting from the beginning. A tape can also have large tape errors on it, as long as they are contained within a stream. A seek to the next tape marker can be performed and other streams can be read.
As discussed above, the invention provides a redundant system that records simultaneously to both a hard disk 146, 815a, 815b and to a tape 199, 840 in tape drive 144, 822. The hard disks maybe removable, for further redundancy. Further redundancy is provided by semiconductor memory 823 and CD burner 824.
Another factor in the architecture of the present system is also to incorporate a built-in automatic data recovery system such that no extraordinary measures are needed to recover data should there be a massive failure to the system. The first phase of the data recovery system is to store the tape index information redundantly using different technology for each instance as shown in
As mentioned in connection with
There is a dedicated tape partition 1242 for the index. This index mirrors the primary system and is used only in the case of errors in the primary system. Normally, this is only accessed in case of trouble, as it does introduce a delay and tape re-positioning not present in the primary system.
There is also a partition directory copy 1246 at the end of data for each partition. Once a video segment is created, and before the tape is re-positioned, the tape directory is duplicated at the back end on the current partition. This directory is only valid for all data previous to this position. However, it affords another level of redundancy available if any of the previous systems have a problem.
A disk index 1244 is also created on originating system, such as 814a, that streams the data to tape. This index is preferably stored on the originating system hard disk such as 815b. This index is a last resort if a tape should have a triple failure. Inserting the tape cassette, such as 840, into its originating video source, such as 822, will cause the system to automatically repair the tape.
The tape drive 822 includes a processor 1520 that executes software 1522. Memory 1524, input/output device 1526, and tape drive 822 are in communication with the processor 1320. The tape drive 1328 is configured to write the video signal 1304 onto a digital tape 1330a. In the preferred embodiment, the tape deck 822 is configured to write set marks 1238 (
The EEPROM preferably hold a compressed directory, and preferably uses set mark or tape marker counts instead of byte counts to indicate correspondence of individual portions of the directory to the tape. Should a catastrophic EEPROM failure or corruption happen, the index can be reconstructed by searching for tape markers at full tape seek speed.
The system architecture revolves around the design goal of minimizing the loss of video. Under any realistic circumstances, the content is recoverable. While the data is being written, is the most sensitive time. Thus, during this time the data is stored on disk, tape, and/or a solid state memory. Once the data is written, it can be duplicated to a library and preserved as needed. In addition, the data is a recoverable by a number of user-serviceable processes, meaning no special recovery software is needed. In the prior art, what the tape has physically recorded and what the system thinks it recorded could be out of sync due to video still in the tape drive cache or pending memory buffers waiting to be transferred, but not physically written. The prior art systems relied on a directory structure that had to be in sync with the data on tape and hard disk. Due to the nature of the surveillance business, it is often the case that the tape video system (TVS) would be put in unstable situations; e.g., when power is suddenly turned off before the directory could be written, when the tape recovery operation is interrupted , etc. For example, anything from a brief power outage to an explosion could put the video data in jeopardy, as everything on tape would be considered lost if the directory was corrupt. The system, in accordance with the principles of the present invention, may perform its resume operation by relying on the total data written to tape, which is easily determined via the partition information, and performing a calculation to determine where the data was interrupted from the disk mirror. This process provides reliability, the blocks written to tape as well as a starting time are known for certain.
In keeping with the absolute reliability goal, the surveillance system according to the invention is designed to automatically correct tape errors and automatically take action to prevent tape defects from corrupting data. In the worst case scenario where data is corrupted on a tape, the tape will automatically reconstruct the data if possible.
The system of the invention provides zero down-time recording. As discussed above, the recording can be reviewed and recorded simultaneously. The writing to disk, the buffers, and the partition information allow the tape to be swapped while the system is being used. The system also includes a software function that permits pre and post event recording for from ten seconds to two hours before and after an event.
The system is capable of producing native MPG format directly, which increases the recovery ability by the user at multiple levels. Should anything happen to the tape, a backup version is available to the user, either an entire tape can be re-created from scratch, or the video can be off-loaded from the tape video system via many methods either via a network or locally.
The system incorporates true embedded multi-channel recordings. Instead of using one MPG per channel, all channels maybe merged into a single MPG file. This enables synchronization between channels, ease of editing/clipping, and ease of maintaining archives of video, since each channel maybe integrated into a single file.
Via the abstract encoder and decoder systems, the system provides native support For MPEG-4 Part 10 (H.264) and MPEG-4. The system is designed to handle newer compression schemes without change to the recording pipeline or format of the tape. As a result, the system is capable of recording any MPG standard compression scheme.
In addition to handling a variety of compression schemes, it is not necessary for all channels to have the same bit rate, or, for that matter, even use the same compression scheme. In one embodiment, one channel maybe configured for a MPEG-2 hi-bit rate and another channel for H.264 low bit rate. This allows for flexibility to choose how to allocate total bandwidth instead of dividing evenly between video signals. Thus, a primary camera can be given a bit rate that has the highest quality, and the back-up cameras can share a lower bit rate, without sacrificing the video quality of the primary camera.
The design of the system has no upper limit imposed on the number of true independent channels it can handle. There is a practical limit based on recording time and capacity of storage, but other than that, there is no limit. For example, eight or more channels are easily achievable.
The system includes built-in full SSL security Web server technology. The system preferably includes the ability to web-enable any capture box, to allow for web access of the video and/or control. Users can remotely monitor their tape video system from anywhere in the world. In addition, no unauthorized access is possible. In one embodiment, the same security features that PayPal™ uses to secure money transactions to millions of people is utilized to secure content stored in the tape video system. Additional and/or alternative security features may be utilized to maintain unauthorized access to the tape video system. The tape video system includes wireless access, thereby allowing mobile systems to be monitored by either pulling up to a designated monitoring bay or even using a simple Pocket PC to monitor it.
A user interface (“Command Center”) maybe completely web-based, thereby giving the user flexibility in accessing content stored on the tape video system via web-based playback and monitoring. Because the system may be web-based, users do not need a copy of the application resident on a local computer to access and utilize the system. A user may simply logon to the system's TVS box with Internet Explorer or other web application and the web page presented is the Command Center. No sacrifice has been made to the functionality for this. The web-based technology will pass through firewalls; in fact, it acts just like a normal web page. It allows users to run on or off site. If a user prefers, the user can visit a system website and run the user interface from there. The web-based user interface accesses only the customer's local files for tape and disk playback Because the Command Center is completely web-based, a custom functionality and look can be done quite easily.
The tape video system makes a great core component for many other video-based applications. To leverage this even further, a tape video system can be fully controlled and video manipulated via a built-in PHP server-side script engine as understood in the art. Any type of application that can be envisioned can be scripted directly on a tape video system without changing the application. The system provides fully database-driven video playback and multiple tape video system servers networked together. Because of the industry standardized PHP engine, many existing PHP applications can be run directly on the tape video system. In combination with the built-in web-based server, the system's networking ability, and its built-in telemetry clock, it is possible to create huge synchronized capture arrays for such things as stadiums, football fields, casinos, or street traffic. There is no upper performance limit in this case, and a 100-camera system, all monitored by a single administrator either on or off site anywhere in the world, is possible.
The tape video system incorporates a built-in telemetry system. Not only does the system record the video/audio, but it also embeds telemetry into the stream, the location of the system, the elevation, and even the speed of travel of the capture system, among other parameters. This is valuable for indisputable court evidence.
The tape video system provides the highest quality video possible with today's technology. It can handle the highest amount of full D1 resolution cameras, the highest quality, the highest recording capacity, and the highest specifications for mission critical applications in the industry. The system, according to the principles of the present invention, embodies several innovations for the implementation of ultra-large real-time recording and playback of video to cartridge tape as a completely digital process. These innovations permit streaming media to/from tape in a manner analogous to a recordable DVD. The advantages of streaming media to/from digital tape over other digital methods (such as DVD, CD-ROM, or even a hard disk) include higher capacity and resolution, increased durability, and additional fault-tolerant capability.
Table B below compares the present system of streaming tape utilizing the principles of the present invention with other conventional recording media. In this table, “Shock Resistant” means that the system can record/play without trouble during a slight shock such as is common in mobile environments. Serious Error Recovery means that the system can recover from a serious permanent error, during recording or playback. Removable Media includes that the recording medium is removable, economical, and replaceable. DVD and CD-ROM are recordable but limited in that recording is a once-only operation, and is not capable of start-stop recording. While a hard disk can handle moderate shocks, but will be destroyed in a removable application if dropped. Although analog tape will continue to record during a shock, it will produce many undesirable artifacts for several seconds after the initial shock.
The only other media having competing capacity to the system of the invention is a high-capacity hard drive. However, such a large hard drive is not practical for removable media applications from a cost standpoint. The other possible removable media types, such as CD-ROM, DVD, HD-DVD, and Blue-Ray, lack any real capacity comparable to the present streaming tape system. In addition, they are totally unusable for recording in a mobile environment, since the slightest shock can render the entire recording unusable. None of these removable media types can withstand bumps and shocks while recording and playing back with no artifacts present. In addition, these removable media types are not typically recordable; and, if they are, they possess the ability to record once, as in the case of DVD or higher.
Other features by using digital tape for the removable media in a surveillance system include: the system can record multi-channel synchronized video (multiple camera recordings at once); the system can record to inexpensive, convenient, rugged cartridge tape, with many more hours of recording than all the existing and future planned streaming devices; excellence in recording and playback of HDTV resolution video; cartridge tape is more rugged and shock resistant than all other forms of storage; and ability to integrate additional digital information other than video, such as telemetry, roster information, subtitling, on-screen display information, synchronization information, etc. while still maintaining high quality.
The system can record up to broadcast-quality video from up to 16 cameras attached to a single recording device. In addition to traditional video, the system also records audio and GPS information to authenticate the exact time, date, and location of events, providing the ultimate solution for surveillance applications. Video is recorded to standards-based video formats that can be played back on any standard PC, which provides flexibility and interoperability when managing or reconstructing incidents.
While virtually all solutions currently available in the mobile digital video surveillance market are disk-based, the present system uses both hard disk and removable digital tape storage to provide critical backup support and an economic advantage. The system also provides a method of archiving large video files to removable digital tape that are consistent with broadcast quality MPEG-2 DVD.
The system according to the invention provides ease of use and flexibility, minimal downtime, and multi-partitioning. It is also an economical solution with a low cost-per-gigabyte, while delivering superior performance, density, and reliability.
The system is configured to provide small form-factor, high capacity, and reliability needed for demanding security applications at a reasonable cost. The system may further be configured to provide a full-motion, high-resolution video surveillance system with a highly reliable, removable storage solution to manage mission-critical needs.
The tape drives of the system may include helical-scan recording, highly durable advanced metal evaporated (AME) media formulation, and a performance enhancing memory-in-cassette chip. One embodiment of the system features a range of capacities and performance solutions up to 200 GB native storage capacity and sustained native transfer rates of up to 24 MB/second.
The system is operable to provide zero downtime recording using broadcast-quality video with four and a half times the resolution of other digital systems, making it superior to other digital or analog security systems on the market today.
The system may also be configured to provide a true 24/7 recording of over 240 hours of continuous broadcast-quality video on Sony AIT-3 digital tape with no manual intervention. The Sony Advanced Intelligent Tape™ (AIT) platform was selected as the removable digital tape storage technology because it provides high-capacity storage for data security and archiving, high-speed file location and file access, backward read and write compatibility, and write once, read many (WORM) functionality.
Although the system described herein is directed to surveillance systems, it should be understood that the principles of the present invention could be applied to non-surveillance systems. For example, because the system is configured to handle video streaming, it should be understood that the system could be used with movies or other video recordings. Further, the same principles could be applied to audio signals or other continuous streamed digital information that would benefit from the use of large storage media with high-speed searching capabilities, including telemetry and other recording systems.
It should be understood that the particular embodiments shown in the drawings and described within this specification are for purposes of example and should not be construed to limit the invention, which will be described in the claims below. For example, although the system is particularly useful if MPEG-2 compression is used, any MPEG compression scheme, known now or in the future, may be substituted. In fact, since this is the first disclosure of any direct streaming of compressed audio/visual signal to digital tape, it is inventive to use any present or future digital compression scheme, for example, MJPEG sometimes referred to as motion JPEG (Joint Photographic Experts Group), and it is contemplated that the future versions of the invention will include hardware and software to utilize many other digital compression apparatuses and processes. Further, it is evident that those skilled in the art may now make numerous uses and modifications of the specific embodiments described without departing from the inventive concepts. It is also evident that the methods recited may in many instances be performed in a different order; or equivalent structures and processes maybe substituted for the various structures and processes described. Consequently, the invention is to be construed as embracing each and every novel feature and novel combination of features present in and/or possessed by the invention herein described.
This Application is a Continuation-in-part of U.S. patent application Ser. No. 10/285862 filed on Nov. 1, 2002, which claims the benefit of U.S. Provisional Patent Application 60/415905 filed on Oct. 3, 2002, and U.S. Provisional Patent Application 60/335926 filed on Nov. 1, 2001. This Application also claims the benefit of U.S. Provisional Patent Application 60/719052 filed on Sep. 20, 2005, and U.S. Provisional Patent Application 60/776804 filed on Feb. 24, 2006. All of the referenced applications are incorporated by reference to the same extent as though fully disclosed herein.
| Number | Date | Country | |
|---|---|---|---|
| 60415905 | Oct 2002 | US | |
| 60335926 | Nov 2001 | US | |
| 60776804 | Feb 2006 | US | |
| 60719052 | Sep 2005 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 10285862 | Nov 2002 | US |
| Child | 11502062 | Aug 2006 | US |
