Patent Application
20140139632
A number of different techniques are known for generating three-dimensional (3D) images of a spatial scene in real time. For example, 3D images of a spatial scene may be generated using triangulation based on multiple two-dimensional (2D) images captured by multiple cameras at different locations. However, a significant drawback of such a technique is that it generally requires very intensive computations, and can therefore consume an excessive amount of the available computational resources of a computer or other processing device. Also, it can be difficult to generate an accurate 3D image under conditions involving insufficient ambient lighting when using such a technique.
Other known techniques include directly generating a 3D image using a depth imager such as a time of flight (ToF) camera or a structured light (SL) camera. Cameras of this type are usually compact, provide rapid image generation, and operate in the near-infrared part of the electromagnetic spectrum. As a result, ToF and SL cameras are commonly used in machine vision applications such as gesture recognition in video gaming systems or other types of image processing systems implementing gesture-based human-machine interfaces. ToF and SL cameras are also utilized in a wide variety of other machine vision applications, including, for example, face detection and singular or multiple person tracking.
A typical conventional ToF camera includes an optical source comprising, for example, one or more light-emitting diodes (LEDs) or laser diodes. Each such LED or laser diode is controlled to produce continuous wave (CW) output light having substantially constant amplitude and frequency. The output light illuminates a scene to be imaged and is scattered or reflected by objects in the scene. The resulting return light is detected and utilized to create a depth map or other type of 3D image. This more particularly involves, for example, utilizing phase differences between the output light and the return light to determine distances to the objects in the scene. Also, the amplitude of the return light is used to determine intensity levels for the image.
A typical conventional SL camera includes an optical source comprising, for example, a laser and an associated mechanical laser scanning system. Although the laser is mechanically scanned in the SL camera, it nonetheless produces output light having substantially constant amplitude. However, the output light from the SL camera is not modulated at any particular frequency as is the CW output light from a ToF camera. The laser and mechanical laser scanning system are part of a stripe projector of the SL camera that is configured to project narrow stripes of light onto the surface of objects in a scene. This produces lines of illumination that appear distorted at a detector array of the SL camera because the projector and the detector array have different perspectives of the objects. A triangulation approach is used to determine an exact geometric reconstruction of object surface shape.
Both ToF and SL cameras generally operate with uniform illumination of a rectangular field of view (FoV). Moreover, as indicated above, the output light produced by a ToF camera has substantially constant amplitude and frequency, and the output light produced by an SL camera has substantially constant amplitude.
In one embodiment, a depth imager is configured to capture a first frame of a scene using illumination of a first type, to define a first area associated with an object of interest in the first frame, to identify a second area to be adaptively illuminated based on expected movement of the object of interest, to capture a second frame of the scene with adaptive illumination of the second area using illumination of a second type different than the first type, and to attempt to detect the object of interest in the second frame.
The illumination of the first type may comprise, for example, substantially uniform illumination over a designated field of view, and the illumination of the second type may comprise illumination of substantially only the second area. Numerous other illumination types may be used.
Other embodiments of the invention include but are not limited to methods, systems, integrated circuits, and computer-readable media storing program code which when executed causes a processing device to perform a method.
Embodiments of the invention will be illustrated herein in conjunction with exemplary image processing systems that include depth imagers having functionality for adaptive illumination of an object of interest. By way of example, certain embodiments comprise depth imagers such as ToF cameras and SL cameras that are configured to provide adaptive illumination of an object of interest. Such adaptive illumination may include, again by way of example, variations in both output light amplitude and frequency for a ToF camera, or variations in output light amplitude for an SL camera. It should be understood, however, that embodiments of the invention are more generally applicable to any image processing system or associated depth imager in which it is desirable to provide improved detection of objects in depth maps or other types of 3D images.
Although shown as being separate from the processing devices 102 in the present embodiment, the depth imager 101 may be at least partially combined with one or more of the processing devices. Thus, for example, the depth imager 101 may be implemented at least in part using a given one of the processing devices 102. By way of example, a computer may be configured to incorporate depth imager 101.
In a given embodiment, the image processing system 100 is implemented as a video gaming system or other type of gesture-based system that generates images in order to recognize user gestures. The disclosed imaging techniques can be similarly adapted for use in a wide variety of other systems requiring a gesture-based human-machine interface, and can also be applied to numerous applications other than gesture recognition, such as machine vision systems involving face detection, person tracking or other techniques that process depth images from a depth imager.
The depth imager 101 as shown in
The control circuitry 105 comprises driver circuits for the optical sources 106. Each of the optical sources may have an associated driver circuit, or multiple optical sources may share a common driver circuit. Examples of driver circuits suitable for use in embodiments of the present invention are disclosed in U.S. patent application Ser. No. 13/658,153, filed Oct. 23, 2012 and entitled “Optical Source Driver Circuit for Depth Imager,” which is commonly assigned herewith and incorporated by reference herein.
The control circuitry 105 controls the optical sources 106 so as to generate output light having particular characteristics. Ramped and stepped examples of output light amplitude and frequency variations that may be provided utilizing a given driver circuit of the control circuitry 105 in a depth imager comprising a ToF camera can be found in the above-cited U.S. patent application Ser. No. 13/658,153. The output light illuminates a scene to be imaged and the resulting return light is detected using detector arrays 108 and then further processed in control circuitry 105 and other components of depth imager 101 in order to create a depth map or other type of 3D image.
The driver circuits of control circuitry 105 can therefore be configured to generate driver signals having designated types of amplitude and frequency variations, in a manner that provides significantly improved performance in depth imager 101 relative to conventional depth imagers. For example, such an arrangement may be configured to allow particularly efficient optimization of not only driver signal amplitude and frequency, but also other parameters such as an integration time window.
The depth imager 101 in the present embodiment is assumed to be implemented using at least one processing device and comprises a processor 110 coupled to a memory 112. The processor 110 executes software code stored in the memory 112 in order to direct at least a portion of the operation of the optical sources 106 and the detector arrays 108 via the control circuitry 105. The depth imager 101 also comprises a network interface 114 that supports communication over network 104.
The processor 110 may comprise, for example, a microprocessor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a central processing unit (CPU), an arithmetic logic unit (ALU), a digital signal processor (DSP), or other similar processing device component, as well as other types and arrangements of image processing circuitry, in any combination.
The memory 112 stores software code for execution by the processor 110 in implementing portions of the functionality of depth imager 101, such as portions of modules 120, 122, 124, 126, 128 and 130 to be described below. A given such memory that stores software code for execution by a corresponding processor is an example of what is more generally referred to herein as a computer-readable medium or other type of computer program product having computer program code embodied therein, and may comprise, for example, electronic memory such as random access memory (RAM) or read-only memory (ROM), magnetic memory, optical memory, or other types of storage devices in any combination. As indicated above, the processor may comprise portions or combinations of a microprocessor, ASIC, FPGA, CPU, ALU, DSP or other image processing circuitry.
It should therefore be appreciated that embodiments of the invention may be implemented in the form of integrated circuits. In a given such integrated circuit implementation, identical die are typically formed in a repeated pattern on a surface of a semiconductor wafer. Each die includes, for example, at least a portion of control circuitry 105 and possibly other image processing circuitry of depth imager 101 as described herein, and may further include other structures or circuits. The individual die are cut or diced from the wafer, then packaged as an integrated circuit. One skilled in the art would know how to dice wafers and package die to produce integrated circuits. Integrated circuits so manufactured are considered embodiments of the invention.
The network 104 may comprise a wide area network (WAN) such as the Internet, a local area network (LAN), a cellular network, or any other type of network, as well as combinations of multiple networks. The network interface 114 of the depth imager 101 may comprise one or more conventional transceivers or other network interface circuitry configured to allow the depth imager 101 to communicate over network 104 with similar network interfaces in each of the processing devices 102.
The depth imager 101 in the present embodiment is generally configured to capture a first frame of a scene using illumination of a first type, to define a first area associated with an object of interest in the first frame, to identify a second area to be adaptively illuminated based on expected movement of the object of interest, to capture a second frame of the scene with adaptive illumination of the second area using illumination of a second type different than the first type, and to attempt to detect the object of interest in the second frame.
A given such process may be repeated for one or more additional frames. For example, if the object of interest is detected in the second frame, the process may be repeated for each of one or more additional frames until the object of interest is no longer detected. Thus, the object of interest can be tracked through multiple frames using the depth imager 101 in the present embodiment.
Both the illumination of the first type and the illumination of the second type in the exemplary process described above are generated by the optical sources 106. The illumination of the first type may comprise substantially uniform illumination over a designated field of view, and the illumination of the second type may comprise illumination of substantially only the second area, although other illumination types may be used in other embodiments.
The illumination of the second type may exhibit at least one of a different amplitude and a different frequency relative to the illumination of the first type. For example, in some embodiments, such as one or more ToF camera embodiments, the illumination of the first type comprises optical source output light having a first amplitude and varying in accordance with a first frequency and the illumination of the second type comprises optical source output light having a second amplitude different than the first amplitude and varying in accordance with a second frequency different than the first frequency.
More detailed examples of the above-noted process will be described below in conjunction with the flow diagrams of
For example, a driver circuit of control circuitry 105 in a given embodiment may comprise amplitude control module 134, such that a driver signal provided to at least one of the optical sources 106 varies in amplitude under control of the amplitude control module 134 in accordance with a designated type of amplitude variation, such as a ramped or stepped amplitude variation.
The ramped or stepped amplitude variation can be configured to provide, for example, an increasing amplitude as a function of time, a decreasing amplitude as a function of time, or combinations of increasing and decreasing amplitude. Also, the increasing or decreasing amplitude may follow a linear function or a non-linear function, or combinations of linear and non-linear functions.
In an embodiment with ramped amplitude variation, the amplitude control module 134 may be configured to permit user selection of one or more parameters of the ramped amplitude variation including one or more of a start amplitude, an end amplitude, a bias amplitude and a duration for the ramped amplitude variation.
Similarly, in an embodiment with stepped amplitude variation, the amplitude control module 134 may be configured to permit user selection of one or more parameters of the stepped amplitude variation including a one or more of a start amplitude, an end amplitude, a bias amplitude, an amplitude step size, a time step size and a duration for the stepped amplitude variation.
A driver circuit of control circuitry 105 in a given embodiment may additionally or alternatively comprise frequency control module 136, such that a driver signal provided to at least one of the optical sources 106 varies in frequency under control of the frequency control module 136 in accordance with a designated type of frequency variation, such as a ramped or stepped frequency variation.
The ramped or stepped frequency variation can be configured to provide, for example, an increasing frequency as a function of time, a decreasing frequency as a function of time, or combinations of increasing and decreasing frequency. Also, the increasing or decreasing frequency may follow a linear function or a non-linear function, or combinations of linear and non-linear functions. Moreover, the frequency variations may be synchronized with the previously-described amplitude variations if the driver circuit includes both amplitude control module 134 and frequency control module 136.
In an embodiment with ramped frequency variation, a frequency control module 136 may be configured to permit user selection of one or more parameters of the ramped frequency variation including one or more of a start frequency, an end frequency and a duration for the ramped frequency variation.
Similarly, in an embodiment with stepped frequency variation, the frequency control module 136 may be configured to permit user selection of one or more parameters of the stepped frequency variation including one or more of a start frequency, an end frequency, a frequency step size, a time step size and a duration for the stepped frequency variation.
A wide variety of different types and combinations of amplitude and frequency variations may be used in other embodiments, including variations following linear, exponential, quadratic or arbitrary functions.
It should be noted that the amplitude and frequency control modules 134 and 136 are utilized in an embodiment of depth imager 101 in which amplitude and frequency of output light can be varied, such as a ToF camera.
Other embodiments of depth imager 101 may include, for example, an SL camera in which the output light frequency is generally not varied. In such embodiments, the LUT 132 may comprise an amplitude-only LUT, and the frequency control module 136 may be eliminated, such that only the amplitude of the output light is varied using amplitude control module 134.
Numerous different control module configurations may be used in depth imager 101 to establish different amplitude and frequency variations for a given driver signal waveform. For example, static amplitude and frequency control modules may be used, in which the respective amplitude and frequency variations are not dynamically variable by user selection in conjunction with operation of the depth imager 101 but are instead fixed to particular configurations by design.
Thus, for example, a particular type of amplitude variation and a particular type of frequency variation may be predetermined during a design phase and those predetermined variations may be made fixed rather than variable in the depth imager. Static circuitry arrangements of this type providing at least one of amplitude variation and frequency variation for an optical source driver signal of a depth imager are considered examples of “control modules” as that term is broadly utilized herein, and are distinct from conventional arrangements such as ToF cameras that generally utilize CW output light having substantially constant amplitude and frequency.
As indicated above, the depth imager 101 comprises a plurality of modules 120 through 130 that are utilized in implementing image processing operations of the type mentioned above and utilized in the
Also included in the depth imager 101 in the present embodiment is a parameter optimization module 130 that is illustratively configured to optimize the integration time window of the depth imager 101 as well as optimization of the amplitude and frequency variations provided by respective amplitude and frequency control modules 134 and 136 for a given imaging operation performed by the depth imager 101. For example, the parameter optimization module 130 may be configured to determine an appropriate set of parameters including integration time window, amplitude variation and frequency variation for the given imaging operation.
Such an arrangement allows the depth imager 101 to be configured for optimal performance under a wide variety of different operating conditions, such as distance to objects in the scene, number and type of objects in the scene, and so on. Thus, for example, integration time window length of the depth imager 101 in the present embodiment can be determined in conjunction with selection of driver signal amplitude and frequency variations in a manner that optimizes overall performance under particular conditions.
The parameter optimization module 130 may also be implemented at least in part in the form of software stored in memory 112 and executed by processor 110. It should be noted that terms such as “optimal” and “optimization” as used in this context are intended to be broadly construed, and do not require minimization or maximization of any particular performance measure.
The particular configuration of image processing system 100 as shown in
The operation of the depth imager 101 in various embodiments will now be described in more detail with reference to
In the embodiment to be described in conjunction with
Referring now to
The object of interest is detected and tracked in these multiple frames using the process illustrated by the flow diagram of
In step 300, the first frame including the object of interest is captured with uniform illumination. This uniform illumination may comprise substantially uniform illumination over a designated field of view, and is an example of what is more generally referred to herein as illumination of a first type.
In step 302, the object of interest is detected in the first frame using object detection module 126 and predefined object templates or other information characterizing typical objects of interest as stored in the objects library 122. The detection process may involve, for example, comparing various identified portions of the frame with sets of predefined object templates from the objects library 122.
In step 304, a first area associated with the object of interest in the first frame is defined, using area definition module 124. An example of the first area defined in step 304 may be considered the area identified by multiple+marks in
In step 306, a second area to be adaptively illuminated in the next frame is calculated based on expected movement of the object of interest from frame to frame, also using area definition module 124. Thus, definition of the second area in step 306 takes into account object movement from frame to frame, considering factors such as, for example, speed, acceleration, and direction of movement.
In a given embodiment, this area definition may more particularly involve contour motion prediction based on position as well as speed and linear acceleration in multiple in-plane and out-of-plane directions. The resulting area definition may be characterized not only by a contour but also by an associated epsilon neighborhood. Motion prediction algorithms of this type and suitable for use in embodiments of the invention are well-known to those skilled in the art, and therefore not described in further detail herein.
Also, different types of area definitions may be used for different types of depth imagers. For example, area definition may be based on pixel blocks for a ToF camera and on contours and epsilon neighborhoods for an SL camera.
In step 308, the next frame is captured using adaptive illumination. This frame is the second frame in a first pass through the steps of the process. In the present embodiment, adaptive illumination may be implemented as illumination of substantially only the second area determined in step 306. This is an example of what is more generally referred to herein as illumination of a second type. The adaptive illumination applied in step 308 in the present embodiment may have the same amplitude and frequency as the substantially uniform illumination applied in step 300, but is adaptive in the sense that it is applied to only the second area rather than to the entire field of view. In the embodiment to be described in conjunction with
In adaptively illuminating only a portion of a field of view of a depth imager comprising a ToF camera, certain LEDs in an optical source comprising an LED array of the ToF camera may be turned off. In the case of a depth imager comprising an SL camera, the illuminated portion of the field of view may be adjusted by controlling the scanning range of the mechanical laser scanning system.
In step 310, a determination is made as to whether or not an attempt to detect the object of interest in the second frame has been successful. If the object of interest is detected in the second frame, steps 304, 306 and 308 are repeated for one or more additional frames, until the object of interest is no longer detected. Thus, the
As noted above, it is also possible that the adaptive illumination will involve varying at least one of the amplitude and frequency of the output of the depth imager 101 using the respective amplitude and frequency control modules 134 and 136. Such variations may be particularly useful in situations such as that illustrated in
The object of interest is detected and tracked in these multiple frames using the process illustrated by the flow diagram of
In step 500, the first frame including the object of interest is captured with the initial illumination. This initial illumination has amplitude A0 and frequency F0 and is applied over a designated field of view, and is another example of what is more generally referred to herein as illumination of a first type.
In step 502, the object of interest is detected in the first frame using object detection module 126 and predefined object templates or other information characterizing typical objects of interest as stored in the objects library 122. The detection process may involve, for example, comparing various identified portions of the frame with sets of predefined object templates from the objects library 122.
In step 504, a first area associated with the object of interest in the first frame is defined, using area definition module 124. An example of the first area defined in step 504 may be considered the area identified by multiple+marks in
In step 506, a second area to be adaptively illuminated in the next frame is calculated based on expected movement of the object of interest from frame to frame, also using area definition module 124. As in the
In step 508, the next frame is captured using adaptive illumination having the updated amplitude Ai and frequency Fi. This frame is the second frame in a first pass through the steps of the process. In the present embodiment, adaptive illumination may be implemented as illumination of substantially only the second area determined in step 506. This is another example of what is more generally referred to herein as illumination of a second type. As indicated above, the adaptive illumination applied in step 508 in the present embodiment has different amplitude and frequency value than the initial illumination applied in step 500. It is also adaptive in the sense that it is applied to only the second area rather than to the entire field of view.
In step 510, a determination is made as to whether or not an attempt to detect the object of interest in the second frame has been successful. If the object of interest is detected in the second frame, steps 504, 506 and 508 are repeated for one or more additional frames, until the object of interest is no longer detected. For each such iteration, different amplitude and frequency values may be determined for the adaptive illumination. Thus, the
By way of example, in the
With regard to the amplitude variation, the first amplitude is typically greater than the second amplitude if the expected movement of the object of interest is towards the depth imager, and the first amplitude is typically less than the second amplitude if the expected movement is away from the depth imager. Also, the first amplitude is typically greater than the second amplitude if the expected movement is towards a center of the scene, and the first amplitude is typically less than the second amplitude if the expected movement is away from a center of the scene.
With regard to the frequency variation, the first frequency is typically less than the second frequency if the expected movement is towards the depth imager, and the first frequency is typically greater than the second frequency if the expected movement is away from the depth imager.
As mentioned previously, the amplitude variations may be synchronized with the frequency variations, via appropriate configuration of the amplitude and frequency LUT 132. However, other embodiments may utilize only frequency variations or only amplitude variations. For example, use of ramped or stepped frequency with constant amplitude may be beneficial in cases in which the scene to be imaged comprises multiple objects located at different distances from the depth imager.
As another example, use of ramped or stepped amplitude with constant frequency may be beneficial in cases in which the scene to be imaged comprises a single primary object that is moving either toward or away from the depth imager, or moving from a periphery of the scene to a center of the scene or vice versa. In such arrangements, a decreasing amplitude is expected to be well suited for cases in which the primary object is moving toward the depth imager or from the periphery to the center, and an increasing amplitude is expected to be well suited for cases in which the primary object is moving away from the depth imager or from the center to the periphery.
The amplitude and frequency variations in the embodiment of
It is to be appreciated that the particular processes illustrated in
It should again be emphasized that the embodiments of the invention as described herein are intended to be illustrative only. For example, other embodiments of the invention can be implemented utilizing a wide variety of different types and arrangements of image processing systems, depth imagers, image processing circuitry, control circuitry, modules, processing devices and processing operations than those utilized in the particular embodiments described herein. In addition, the particular assumptions made herein in the context of describing certain embodiments need not apply in other embodiments. These and numerous other alternative embodiments within the scope of the following claims will be readily apparent to those skilled in the art.
