The subject matter disclosed herein relates to a method and apparatus for implementing peer to peer content delivery networks. More specifically, the subject matter pertains to a computer implemented method and apparatus for implementing peer to peer content delivery networks by delaying content delivery to facilitate client use of peer content delivery.
Video streaming allows video content to be delivered to a video player via the Internet. The video content is a video signal generated by a content provider for distribution to video consumers. The video signal may be provided in an uncompressed file format, such as a serial digital interface (SDI) format, or in a compressed format, such as a Moving Picture Experts Group (MPEG) file format or Transport Stream (TS) file format. The video signal is sent to an encoder which converts the file into a live streaming signal. The live streaming signal is preferably a segmented data stream that may be transmitted using standard Hypertext Transport Protocol (HTTP) over the internet. The live streaming signal may include multiple streams, where each stream may have a different data rate and/or different resolution.
Two common formats for the live streaming signal include HTTP Live Streaming (HLS) implemented by Apple® and MPEG-Dynamic Adaptive bitrate Streaming over HTTP (MPEG-DASH) implemented, for example, by web browsers such as Chrome®, Microsoft Edge®, and Firefox®. In addition to the segmented data stream, the encoder generates a manifest file. The manifest file contains information for a video player to play the segmented data stream such as the data rate and resolution of each stream and a playlist providing an address from which the video content may be retrieved. Historically, the encoder has generated a single manifest file for each encoded video signal, where the manifest file is distributed along with the streaming signal.
The live streaming signal and the manifest file are stored in one or more Content Delivery Networks (CDN). Each CDN includes a number of edge servers which store the streaming signal and manifest file until requested by a video player. When the streaming signal is provided to multiple CDNs, the CDNs may be in different geographic locations, such as the West Coast, East Coast, or Midwest. Each video player may select a CDN based, among other things, on its geographic proximity in order to reduce transmission latencies.
The video player may be any suitable electronic device to receive the streaming signal such as a desktop computer, a television, a laptop computer, a tablet, or a mobile phone. A user initiates a request to view desired video content on the video player. The video player includes video management software executing on the video player which has knowledge of the address of the CDN and which may provide a list of video content stored on the CDN to the user. After the user has selected a desired video, the video player, in turn, requests that the video content be transmitted from the CDN.
The streaming video content is commonly encrypted prior to transmission. A standard encryption protocol, such as the Advanced Encryption Standard (AES) may be used. During encoding of the video signal, the encoder may communicate with a key server to obtain an encryption key for the video content and encrypt the video signal as part of the encoding process. The encoder may include the location of the encryption key in the manifest file for the encoded video content and provide the manifest file to the CDN. The manifest file is delivered to the video player when the video content is requested. The video player reads the location of the encryption key from the manifest file, retrieves the encryption key, and decrypts the video content on the player prior to displaying the video content to the user.
Live encoders or video on demand encoders generate video segment files and upload them to origin servers or content delivery networks. Content delivery networks distribute these video segments to their edge servers when users submit requests to play the segments. However, it is common in live stream that every edge node and every user attempts to obtain the latest video segments often in temporal proximity. This causes a significant load on a content delivery network's edge servers and can create bottlenecks for video segment delivery.
What is needed is a method of implementing peer to peer content delivery networks to reduce the load on the content delivery networks and edge servers. What is further needed is such a method and apparatus for implementing peer to peer content delivery networks without extensive modification to video player software.
The subject matter disclosed herein describes a method for implementing peer to peer content delivery networks for video/media segment delivery by deliberately delaying a second client's live stream to serve that client from a first client that previously downloaded the video/media segment from a traditional content delivery network. The method may include assembling a list of available sources for live video segments and their locations based on previously received segment requests and providing customized manifests identifying those available sources from the list as potential peer content delivery networks.
In one embodiment of the disclosure, a computer implemented method embodied in a computer program stored in non-transient memory and executed by a computer processor for providing peer to peer content delivery of a live video stream includes receiving a first request for video content delivery from a first video player, receiving a second request for the video content delivery from a second video player, providing video content to the first video player from a content delivery network, and providing the video content to the second video player from the first video player.
According to another aspect of the disclosure, providing video content to the first video player from a content delivery network includes providing the video content based on a manifest file identifying the content delivery network as a delivery source. Further, providing video content to the second video player from the first video player includes providing the video content based on a manifest file identifying the first video player as a delivery source.
In another aspect of the disclosure, providing the video content to the second video player from the first video player is delayed for a defined time period. In one another aspect, the defined time period may be at least one video segment length+epsilon.
According to another embodiment of the disclosure, a computer implemented system for providing peer to peer content delivery includes a manifest server operative to generate a unique manifest file per request for a video file stored on a content delivery network. A first unique manifest file for a first video player is configured for providing video content to the first video player from a content delivery network and a second unique manifest file for a second video player is configured for providing video content to the second video player from the first video player.
According to another embodiment of the disclosure, a computer implemented method embodied in a computer program stored in non-transient memory and executed by a computer processor for providing peer to peer content delivery of a live video stream is described. The method includes receiving a first request for video content delivery from a first video player, providing video content to the first video player from a content delivery network, receiving a confirmation of receipt and an indication of availability from the first video player, receiving a second request for the video content delivery from a second video player, and providing the video content to the second video player from the first video player.
In one aspect of the disclosure, the confirmation of receipt and indication of availability are received in the defined time period and the video content is provided from the first video player based on the received confirmation and indication. In another aspect of the disclosure, the confirmation of receipt and indication of availability are received after the defined time period and the video content is provided from the content delivery network.
These and other objects, advantages, and features of the disclosure will become apparent to those skilled in the art from the detailed description and the accompanying drawings. It should be understood, however, that the detailed description and accompanying drawings, while indicating preferred embodiments of the present disclosure, are given by way of illustration and not of limitation. Many changes and modifications may be made within the scope of the present disclosure without departing from the spirit thereof, and the disclosure includes all such modifications.
Various exemplary embodiments of the subject matter disclosed herein are illustrated in the accompanying drawings in which like reference numerals represent like parts throughout, and in which:
In describing the preferred embodiments of the disclosure which are illustrated in the drawings, specific terminology will be resorted to for the sake of clarity. However, it is not intended that the disclosure be limited to the specific terms so selected and it is understood that each specific term includes all technical equivalents which operate in a similar manner to accomplish a similar purpose. For example, the word “connected,” “attached,” or terms similar thereto are often used. They are not limited to direct connection but include connection through other elements where such connection is recognized as being equivalent by those skilled in the art.
The various features and advantageous details of the subject matter disclosed herein are explained more fully with reference to the non-limiting embodiments described in detail in the following description.
Turning initially to
In addition to the segmented data stream, the encoder 114 generates a manifest file. The manifest file contains information for a video player 122 to play the segmented data stream such as the data rate and resolution of each stream and a playlist providing an address from which the video content may be retrieved. The encoder 114 generates a single manifest file for each encoded video signal, where the manifest file is distributed along with the streaming signal 116 to a video origin server 131. The manifest files are updated periodically, for example every 10 seconds in one exemplary embodiment, and new segment files are uploaded to the video origin server 131. According to an alternative embodiment, the encoder 114 may be configured to upload manifest files and segment files directly to CDNs 118.
CDNs 118 periodically access video origin server 131 and upload video segments files such that the video segment files will be stored on one or more CDNs 118. The manifest file may include an address of each CDN such that playback may occur from any of the CDNs 118. Each CDN 118 includes a number of edge servers 120 which store the encoded video signal 116 and manifest file until playback of the video content is requested by a video player 122. Manifest files include two types of playlists, master playlists and rendition playlists.
As further illustrated in
Turning next to
After converting the video signal 112 into segments, the encoder 114 encrypts the video signals 112 to prevent unauthorized viewing of the video content. At block 132, the encoder 114 establishes communication with a key server 126 and requests a key to use for encrypting the segmented video signal 112. The key server 126 returns a key to the encoder 114 as shown in block 134. The key used to encrypt the segmented video signal 112 will be referred to herein as the content encryption key. The encoder 114 may use any suitable encryption protocol, such as the Advanced Encryption Standard (AES), to encrypt the segmented video signal using the content encryption key. The location of the key server and the encryption key used to encrypt the segmented video signal is included in a manifest file. The manifest file and the encrypted video signal are then transmitted to an origin server 131. Manifest files are uploaded periodically and new segment files uploaded to the origin server 131 for distribution to CDN 118. The manifest file and the encrypted video signal may alternatively be uploaded directly from the encoder 114 to one or more CDN 118 for storage in one of the edge servers 120, as shown in block 136. Although the video signal is described herein as being an encrypted video signal, the video signal is not required to be encrypted in alternative embodiments.
Referring now also to
With reference also to
When the video player 122 requests the manifest file from the manifest server 124 a first connection is established between the devices. A session identifier is also generated to identify the connection. The session identifier may be generated by the video player 122 or the manifest server 124. For purposes of illustration, it will be assumed that the session identifier is generated by the video player 122. The session identifier is transmitted to the manifest server 124 by the video player 122 when requesting a manifest file. The location of a playlist for each of the streams is also included in the manifest file. Each line begins with the location of the key server to decrypt the video segment, identifies a particular video segment between 1 and 5 (i.e., the “−1”, “−2”, etc. . . . prior to the .ts file extension), and provides the location of video segment in the CDN 118. The manifest server 124 then requests the manifest file from the CDN 118 at block 142. At block 144, the CDN 118 returns the manifest file to the manifest server 124.
Because the manifest server 124 has established a connection with video player 122, it may customize the manifest file prior to returning the manifest file to the video player 122 and provide a unique manifest file to each video player 122. Without the manifest server 124, the video player 122 retrieves the manifest file directly from the CDN 118 and the content of the manifest file is the same for all users. However, because the manifest server 124 is providing a unique manifest file to each player, the manifest file may include identifying information of the video player 122, the user of the video player, or a combination thereof. Further, the manifest file may be modified to include content specific for the user. The manifest server 124 then transmits the manifest file to the video player 122, as shown at block 145.
The manifest file will have the address of the CDN 118 as containing the segmented video content. Therefore, the video player can then start retrieving the video content from the CDN 118. The video player 122 repeatedly requests the next segment in the playlist from the CDN 118 and the CDN returns the requested segment as shown by blocks 146 and 148. The native video player module 128 then decodes the content from the encrypted video segments and displays the requested video content to the user.
When the second video player 123 requests the manifest file from the manifest server 124 a second connection is established. Similar to the first connection, a session identifier is also generated to identify the second connection. The session identifier may be generated by the second video player 123 or the manifest server 124. For purposes of illustration, it will be assumed that the session identifier is generated by the second video player 123. The session identifier is transmitted to the manifest server 124 by the second video player 123 when requesting a manifest file.
However, in making this second connection, manifest server 124 has knowledge of the first connection already established with the video player 122. Video players 122 may advertise back to manifest server 124 (via video HTTP redirect) when the video segment is located and/or available in the video player 122. HTTP Redirect method allows the manifest server 124 to ensure that the video player 122 has received the previous segment as the video player 122 is requesting the next segment from CDN 118. As soon as this video segment is available, manifest server 124 can record the change to the video player 122's available segments and their statistics, source location, etc. This knowledge may be used to determine whether the video player 122 is a good candidate to be used as a potential peer content delivery source. Using the video player 122's IP address, manifest server 124 can determine the geolocation of the video player 122 which may be used to determine which second video players 123, for example having a similar geolocation, would benefit from being directed to video player 122 as a peer content delivery source.
The manifest server 124 then requests the manifest file from the CDN 118 at block 150 and customizes the manifest file for the second connection based on the knowledge regarding the first connection. The customized manifest file may be configured to identify the first player 122 as the location of the video segments allowing first player 122 to function as a peer to peer content delivery network. The manifest server 124 then transmits the customized manifest file to the second video player 123, as shown at block 153.
The customized manifest file will have the address of the video player 122 as containing the segmented video content. Therefore, the second video player 123 can then start retrieving the video content from the video player 122 as opposed to the CDN 118. The second video player 123 repeatedly requests the next segment in the playlist from the video player 122 and the video player 122 returns the requested segment as shown by blocks 154 and 156. The native video player module 128 then decodes the content from the encrypted video segments and displays the requested video content to the user.
The method and apparatus described above implements peer-to-peer CDN by deliberately delaying a client live stream to serve those clients from peer video segment sources. By leveraging customized manifest delivery, manifest server 124 may further be configured to switch between bad video sources (bad peer networks) to good video sources (good peer networks) based on heuristics related to devices and/or connections to devices. Further, a customized manifest may be configured to include a plurality of video players 122 as potential peer content delivery sources. The additional video players 122 and/or the original CDN 118 may be configured to function as either primary and/or failover stream sources.
According to an exemplary embodiment, manifest server 124 may be configured to track delivery of video segments to a large number of video players 122. Accordingly, manifest server 124 will have a record of which video players 122 have loaded which video segments and their availability such that manifest server 124 can identify these video players as potential peer content delivery sources.
In order to implement using video players 122 as peer content delivery sources, during delivery of live video, in which segment requests may be received from a large number of other video players 122 essentially simultaneously, it may be necessary to delay service to selected video players 122, essentially rendering these video players as second video players 123, which will receive content from the first video players 122. Second video players 123 may be been delayed for at least one video segment length (TS) plus a small time interval such as one (1) second.
Manifest server 124 can be configured to include a CDN 118 as a failover source for each rendition of a manifest file for second video players 123. Primary content sources may be the video players 122. However, if a second video player 123 cannot load the video segment from a first video player 122, the second video player 123 will try the failover rendition. Since the failover rendition is also delivered from manifest server 124 based on the request for failover provision of the video content, manifest server 124 may note the failure to determine whether the video player 122 acting as a peer source may have an issue in delivering video segments. These statistics will be tracked during live stream delivery to distinguish bad versus good peer content delivery sources (video players 122).
Advantageously, once a video player 122 has been identified as a live stream source, second video players 123 will start receiving video segments from the video player 122 as customized manifest files are distributed. Utilization of this system and method increases the number of available sources, many potentially having better download metrics that a nearest CDN. Further, utilization of the system and method distributes video segment delivery loads across all regions and using video player 122 resources that may be underutilized to deliver the live video streams.
Portions of the disclosed embodiment are described herein as being implemented on various physical devices, including, but not limited to, a plurality of video players 122, second video players 123, the manifest server 124, the key server 126, the encoder 114, origin server 131, or the edge server 120 within a CDN 118. It would be understood by one skilled in the art that these devices may include processing devices, such as a single microprocessor, multiple microprocessors, co-processors, application specific integrated circuits (ASICs), or other computing devices operating separately, in tandem, or a combination thereof. Further, each of these devices includes storage which may include transitory storage, non-transitory storage, or a combination thereof. The storage may include memory devices such as random access memory (RAM), read-only memory (ROM), solid state memory, and the like. The storage may further include devices configured to read removable storage medium such as CD-ROMs, DVDs, floppy disks, universal serial bus (USB) devices, memory cards, and the like. The processing devices may be configured to read and execute instructions stored in non-transitory storage to perform various operations in the methods described herein.
It should be understood that the disclosure is not limited in its application to the details of construction and arrangements of the components set forth herein. The disclosure is capable of other embodiments and of being practiced or carried out in various ways. Variations and modifications of the foregoing are within the scope of the present disclosure. It also being understood that the technology disclosed and defined herein extends to all alternative combinations of two or more of the individual features mentioned or evident from the text and/or drawings. All of these different combinations constitute various alternative aspects of the present disclosure. The embodiments described herein explain the best modes known for practicing the disclosure and will enable others skilled in the art to utilize the disclosure.
This application claims priority to U.S. provisional application Ser. No. 62/334,163, filed May 10, 2016 and titled Method and Apparatus for Providing Peer-to-Peer Content Delivery, the entire contents of which is incorporated herein by reference.
Number | Name | Date | Kind |
---|---|---|---|
5513260 | Ryan | Apr 1996 | A |
5577135 | Grajski et al. | Nov 1996 | A |
5659613 | Copeland et al. | Aug 1997 | A |
5668603 | Copeland | Sep 1997 | A |
5721788 | Powell et al. | Feb 1998 | A |
5883959 | Kori | Mar 1999 | A |
5917943 | Washizawa | Jun 1999 | A |
6018374 | Wrobleski | Jan 2000 | A |
6026189 | Greenspan | Feb 2000 | A |
6173275 | Caid et al. | Jan 2001 | B1 |
6373960 | Conover et al. | Apr 2002 | B1 |
6381367 | Ryan | Apr 2002 | B1 |
6404925 | Foote et al. | Jun 2002 | B1 |
6438275 | Martins et al. | Aug 2002 | B1 |
6539395 | Gjerdingen et al. | Mar 2003 | B1 |
6711293 | Lowe | Mar 2004 | B1 |
6774917 | Foote et al. | Aug 2004 | B1 |
6785815 | Serret-Avila et al. | Aug 2004 | B1 |
6937766 | Wilf et al. | Aug 2005 | B1 |
6975746 | Davis et al. | Dec 2005 | B2 |
6975755 | Baumberg | Dec 2005 | B1 |
7043019 | Tehranchi et al. | May 2006 | B2 |
7055169 | Delpuch et al. | May 2006 | B2 |
7167574 | Kim | Jan 2007 | B2 |
7177470 | Jasinschi et al. | Feb 2007 | B2 |
7185201 | Rhoads et al. | Feb 2007 | B2 |
7218754 | Schumann et al. | May 2007 | B2 |
7272240 | Goldberg et al. | Sep 2007 | B2 |
7298930 | Erol et al. | Nov 2007 | B1 |
7325013 | Caruso | Jan 2008 | B2 |
7421376 | Caruso et al. | Sep 2008 | B1 |
8160366 | Nakamura et al. | Apr 2012 | B2 |
8200021 | Iwamoto et al. | Jun 2012 | B2 |
8515212 | Bengio et al. | Aug 2013 | B1 |
8515241 | Forsyth et al. | Aug 2013 | B2 |
8607283 | Civanlar et al. | Dec 2013 | B2 |
8677428 | Lewis et al. | Mar 2014 | B2 |
8887215 | Fisher et al. | Nov 2014 | B2 |
8943215 | Patten | Jan 2015 | B2 |
9015151 | Margulis et al. | Apr 2015 | B1 |
9066115 | Cherry et al. | Jun 2015 | B1 |
9066138 | Kraiman et al. | Jun 2015 | B1 |
9167278 | Sinha et al. | Oct 2015 | B2 |
9491499 | Wagenaar et al. | Nov 2016 | B2 |
9613042 | Joseph et al. | Apr 2017 | B1 |
10104137 | Salomons | Oct 2018 | B2 |
10148716 | Joseph et al. | Dec 2018 | B1 |
20020159641 | Whitney et al. | Oct 2002 | A1 |
20030005454 | Rodriguez et al. | Jan 2003 | A1 |
20030033347 | Bolle et al. | Feb 2003 | A1 |
20030045954 | Weare et al. | Mar 2003 | A1 |
20030151621 | McEvilly et al. | Aug 2003 | A1 |
20030195883 | Mojsilovic et al. | Oct 2003 | A1 |
20030229900 | Reisman | Dec 2003 | A1 |
20040022447 | Mukhopadhyay et al. | Feb 2004 | A1 |
20040085339 | Divakaran et al. | May 2004 | A1 |
20040221237 | Foote et al. | Nov 2004 | A1 |
20040260786 | Barile | Dec 2004 | A1 |
20050125368 | Akahori | Jun 2005 | A1 |
20050251532 | Radhakrishnan et al. | Nov 2005 | A1 |
20050262245 | Menon et al. | Nov 2005 | A1 |
20050289347 | Ovadia | Dec 2005 | A1 |
20060029253 | Pace | Feb 2006 | A1 |
20060101060 | Li et al. | May 2006 | A1 |
20060111801 | Weare et al. | May 2006 | A1 |
20060271973 | Jerding et al. | Nov 2006 | A1 |
20060279628 | Fleming | Dec 2006 | A1 |
20060291690 | Roberts | Dec 2006 | A1 |
20070025606 | Gholap et al. | Feb 2007 | A1 |
20070128899 | Mayer | Jun 2007 | A1 |
20070157231 | Eldering et al. | Jul 2007 | A1 |
20070217676 | Grauman et al. | Sep 2007 | A1 |
20070253594 | Lu et al. | Nov 2007 | A1 |
20070282898 | Stark et al. | Dec 2007 | A1 |
20080027931 | Lu et al. | Jan 2008 | A1 |
20080040807 | Lu et al. | Feb 2008 | A1 |
20080123976 | Coombs et al. | May 2008 | A1 |
20080166057 | Nakajima | Jul 2008 | A1 |
20080186413 | Someya et al. | Aug 2008 | A1 |
20080271080 | Gossweiler et al. | Oct 2008 | A1 |
20090074235 | Lahr et al. | Mar 2009 | A1 |
20090113512 | Collet et al. | Apr 2009 | A1 |
20090193473 | Moon et al. | Jul 2009 | A1 |
20090204901 | Dharmaji et al. | Aug 2009 | A1 |
20090324026 | Kletter | Dec 2009 | A1 |
20100162330 | Herlein et al. | Jun 2010 | A1 |
20100299438 | Zimmerman et al. | Nov 2010 | A1 |
20110078551 | Zhang et al. | Mar 2011 | A1 |
20110191692 | Walsh et al. | Aug 2011 | A1 |
20110219035 | Korsunsky et al. | Sep 2011 | A1 |
20110314493 | Lemire et al. | Dec 2011 | A1 |
20120159337 | Travilla et al. | Jun 2012 | A1 |
20120216121 | Lin et al. | Aug 2012 | A1 |
20120240176 | Ma et al. | Sep 2012 | A1 |
20120275597 | Knox et al. | Nov 2012 | A1 |
20130163758 | Swaminathan et al. | Jun 2013 | A1 |
20130195204 | Reznik et al. | Aug 2013 | A1 |
20130291002 | Rothschild et al. | Oct 2013 | A1 |
20130311780 | Besehanic | Nov 2013 | A1 |
20140095890 | Mangalore et al. | Apr 2014 | A1 |
20140115060 | Kim et al. | Apr 2014 | A1 |
20140201334 | Wang | Jul 2014 | A1 |
20140259051 | Strein et al. | Sep 2014 | A1 |
20140280781 | Gregotski | Sep 2014 | A1 |
20150058709 | Zaletel | Feb 2015 | A1 |
20150074232 | Phillips et al. | Mar 2015 | A1 |
20150113021 | Martin | Apr 2015 | A1 |
20150172342 | Yin | Jun 2015 | A1 |
20150208103 | Guntur | Jul 2015 | A1 |
20150346832 | Cole et al. | Dec 2015 | A1 |
20150365622 | Ushiyama | Dec 2015 | A1 |
20150365725 | Belyaev et al. | Dec 2015 | A1 |
20150382042 | Wagenaar et al. | Dec 2015 | A1 |
20160065946 | Cole et al. | Mar 2016 | A1 |
20160080470 | Shanson | Mar 2016 | A1 |
20160080810 | Dutta et al. | Mar 2016 | A1 |
20160127260 | Gordon | May 2016 | A1 |
20160127440 | Gordon | May 2016 | A1 |
20160134910 | Davis et al. | May 2016 | A1 |
20160149699 | Gauda | May 2016 | A1 |
20160173961 | Coan et al. | Jun 2016 | A1 |
20160198202 | Van Brandenburg et al. | Jul 2016 | A1 |
20160224799 | Uzen et al. | Aug 2016 | A1 |
20160227279 | Fang et al. | Aug 2016 | A1 |
20160316233 | Ghadi et al. | Oct 2016 | A1 |
20160337704 | Binder et al. | Nov 2016 | A1 |
20170064400 | Riegel et al. | Mar 2017 | A1 |
20170070758 | Phillips et al. | Mar 2017 | A1 |
20170085933 | Czeck, Jr. et al. | Mar 2017 | A1 |
20170150190 | Tarbox et al. | May 2017 | A1 |
20170280181 | Ramaley | Sep 2017 | A1 |
20170302753 | Larumbe et al. | Oct 2017 | A1 |
20170308681 | Gould et al. | Oct 2017 | A1 |
20170353516 | Gordon | Dec 2017 | A1 |
20180063594 | Alexander et al. | Mar 2018 | A1 |
Number | Date | Country |
---|---|---|
101325687 | Dec 2008 | CN |
101589617 | Nov 2009 | CN |
101689184 | Mar 2010 | CN |
103491457 | Jan 2014 | CN |
2006122320 | Nov 2006 | WO |
2007148290 | Dec 2007 | WO |
Entry |
---|
D. G. Lowe, “Object Recognition from Local Scale-Invariant Features”, in International Conference on Computer Vision, vol. 2, 1999, pp. 1-8. |
K. Mihcak and R. Venkatesan, “Blind Image Watermarking Via Derivation and Quantization of Robust Semi-Global Statistics”, in IEEE International Conference on Acoustics, Speech and Signal Processing, 2002, (4 pages). |
T. Lindeberg, “Feature Detection with Automatic Scale Selection”, International Journal of Computer Vision, vol. 30, No. 2, 1998, pp. 1-51. |
A. Andoni and P. Indyk, “Near-Optimal Hashing Algorithms for Approximate Nearest Neighbor in High Dimensions”, in Proceedings of the Symposium on Foundations of Computer Science, Jan. 2006, 10 pages. |
A. Joly, O. Buisson and C. Frelicot, “Content-based Copy Retrieval using Distortion-based Probabilistic Similarity Search”, IEEE Transactions on Multimedia, vol. 9, No. 2, p. 1-14, Feb. 2007. |
J. Matas, O. Chum, M. Urban, T. Pajdla, “Robust Wide Baseline Stereo from Maximally Stable Extremal Regions”, in British Machine Vision Conference., 2002, pp. 384-393. |
K. Mikolajczyk and C. Schmid, “Indexing based on scale invariant interest points”, in Proc. ICCV, 2001, 7 pages. |
Chinese Patent Application No. 201780011326.6, Office Action dated Mar. 9, 2020, 6 pages. |
Chinese Patent Application No. 201780011328.5, Office Action dated Mar. 9, 2020, 8 pages. |
Number | Date | Country | |
---|---|---|---|
62334163 | May 2016 | US |