WO2008063614A2 - Method of and apparatus for performing motion-based object extraction and tracking in video - Google Patents

Method of and apparatus for performing motion-based object extraction and tracking in video Download PDF

Info

Publication number
WO2008063614A2
WO2008063614A2 PCT/US2007/024197 US2007024197W WO2008063614A2 WO 2008063614 A2 WO2008063614 A2 WO 2008063614A2 US 2007024197 W US2007024197 W US 2007024197W WO 2008063614 A2 WO2008063614 A2 WO 2008063614A2
Authority
WO
WIPO (PCT)
Prior art keywords
computer
regions
tracking
program code
readable program
Prior art date
Application number
PCT/US2007/024197
Other languages
French (fr)
Other versions
WO2008063614A3 (en
Inventor
Eitan Sharon
Achiezer Brandt
Original Assignee
Rexee, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority to US86655206P priority Critical
Priority to US60/866,552 priority
Priority to US60/869,279 priority
Priority to US86927106P priority
Priority to US86927906P priority
Priority to US60/869,271 priority
Priority to US11/687,261 priority patent/US8379915B2/en
Priority to US11/687,341 priority
Priority to US11/687,300 priority
Priority to US11/687,290 priority
Priority to US11/687,326 priority patent/US20080120291A1/en
Priority to US11/687,326 priority
Priority to US11/687,290 priority patent/US20080120290A1/en
Priority to US11/687,341 priority patent/US8488839B2/en
Priority to US11/687,300 priority patent/US20080120328A1/en
Priority to US11/687,261 priority
Application filed by Rexee, Inc. filed Critical Rexee, Inc.
Publication of WO2008063614A2 publication Critical patent/WO2008063614A2/en
Publication of WO2008063614A3 publication Critical patent/WO2008063614A3/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/20Analysis of motion
    • G06T7/215Motion-based segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/70Information retrieval; Database structures therefor; File system structures therefor of video data
    • G06F16/78Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/783Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • G06F16/7837Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using objects detected or recognised in the video content
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/70Information retrieval; Database structures therefor; File system structures therefor of video data
    • G06F16/78Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/783Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • G06F16/7847Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using low-level visual features of the video content
    • G06F16/786Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using low-level visual features of the video content using motion, e.g. object motion or camera motion
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/00624Recognising scenes, i.e. recognition of a whole field of perception; recognising scene-specific objects
    • G06K9/00711Recognising video content, e.g. extracting audiovisual features from movies, extracting representative key-frames, discriminating news vs. sport content
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10016Video; Image sequence

Abstract

A method of extracting objects from a video includes the identification of regions to be tracked, tracking the regions across several frames, calculating motions of the regions, and identifying new regions to be tracked. Regions of a selected frame of the video that are appropriate for tracking are selected. Tracking of the identified regions is then performed across frames subsequent to the selected frame of the video. Motions of the identified regions are calculated, and new regions for tracking are identified.

Description

Title of the Invention

Method Of and Apparatus For Performing Motion-Based Object Extraction And Tracking In Video

Cross Reference to related applications

This application claims priority of U.S. Provisional Application Nos. 60/869,271 entitled Rexee "Video Aboutness" Search and 60/869,279 entitled Robust Motion Estimation Using the Brandt-Dym Line Averages filed December 8, 2006 and 60/866,552 entitled Apparatus for and Method of Providing Interactive Video Content filed November 20, 2006; U.S. Patent Application Serial No. 1 1/687,290 entitled Apparatus for Performing a Weight- Based Search; Serial No. 11/687,300 entitled Method of Performing a Weight-Based Search; Serial No. 1 1/387,326 entitled Computer Program Implementing a Weight-Based Search; Serial No. 11/687,261 entitled Method of Performing Motion-Based Object Extraction and Tracking in Video; and Serial No. 11/687,341 entitled Computer Program and Apparatus for Motion-Based Object Extraction and Tacking in Video (all filed March 16, 2007), all of the previously cited provisional and non-provisional applications being incorporated herein by reference in their entireties.

Field of the Invention

The invention relates to processing of video to identify salient objects and, in particular, using motion-based analysis to extract and track salient objects in time-space.

Background of the Invention

Videos are composed of series of still images, each representing photometric qualities. To be useful in many applications, there is a need for automated correlation between these photometric features and physically cognizable objects, e.g., people, scenery, etc. Identification of physical objects may be accomplished by object extraction. With the proliferation of videos (e.g., on the Internet), there is an increasing need for efficient methods and apparatus for extracting objects to support object-based tagging and searching of videos. Summary of the Invention

The invention is directed toward a method of performing object extraction from video content and use of such methods and systems to generate indexed video. According to one aspect of the invention, a method of extracting objects from a video comprising steps of identifying regions to be tracked, tracking the regions across several frames, calculating motions of the regions, and identifying new regions to be tracked. Although recited in a particular sequence, the steps and substeps are capable of rearrangement within the scope of the invention. Thus, according to one implementation, regions of a selected frame of the video that are appropriate for tracking are selected. Tracking of the identified regions is performed across frames subsequent to the selected frame of the video. Motions of the identified regions are calculated, and new regions for tracking are identified. Note that the term "video" is used herein in its broadest sense to include an electronically stored sequence of still images that represent scenes in motion and "video content" means image information stored in the form of a video.

According to an aspect of the invention, the step of identifying new regions for tracking may be responsive to the steps of calculating motions of the identified regions and/or tracking identified regions across frames. According to another aspect of the invention, the step of identifying regions of the selected frame may include steps of segmenting at least one frame of the video to partition the frame into coherent regions and/or extracting local motion estimations for various points in the selected frame across the frames subsequent to the selected frame.

According to another aspect of the invention, the step of tracking identified regions across the frames subsequent to the selected frame may include a step of determining a motion of each of the identified regions of the selected frame across the frames subsequent to the selected frame.

According to another aspect of the invention, the step of determining a motion of each of the identified regions may include (i) fitting the identified regions to the frames subsequent to the selected frame and/or (ii) estimating local motion within the selected frame.

According to another aspect of the invention, the step of tracking identified regions across the frames subsequent to the selected frame may include a step of (i) determining whether a moveable object is trackable and/or (ii) determining whether one of the frames subsequent to the selected frame is substantially different from the selected frame.

According to other aspects of the invention, the step of calculating motions of the identified regions may include a step of (i) determining a motion of each of the identified regions of the selected frame across the frames subsequent to the selected frame, (ii) grouping together ones of the identified regions based on their common motion to generate more complex regions of interest, (iii) determining a camera motion and factoring the camera motion out to identify salient objects that move differently from the camera motion, (iv) identifying a salient object and continuing tracking of the salient object, and/or (iv) receiving user input designating a salient object of interest.

According to another aspect of the invention, a method of generating an indexed video includes identifying regions of a frame of the video appropriate for tracking, tracking identified regions across subsequent frames of the video, calculating motions of the identified regions including identifying coherent video objects, identifying new regions for tracking in response to the calculated motions, creating tubes that relate to the coherent video objects, and indexing and storing the tubes in a searchable database.

According to another aspect of the invention, the step of tracking may include generating tracking information and the step of creating tubes includes a step of collecting the tracking information into space-time threads.

According to another aspect of the invention, the step of creating tubes may include steps of identifying common salient objects and joining ones of the tubes of respective ones of the common salient objects.

According to another aspect of the invention, steps of analyzing tube features to identify the weights, tagging and assigning weights to each of the tubes may further be included.

According another aspect of the invention, a computer program includes a computer usable medium having computer readable program code embodied therein for extracting objects from a video. The computer readable program code may include computer readable program code for causing the computer to extract one or more objects from video content and/or to generate indexed video. According to one aspect of the invention, computer readable program code for causing a computer to identify regions to be tracked, track the regions across several frames, calculate motions of the regions, and identify new regions to be tracked. Although recited in a particular sequence, the steps and substeps performed by the computer program code are capable of rearrangement within the scope of the invention. Thus, according to one implementation, regions of a selected frame of the video that are appropriate for tracking are selected. Tracking of the identified regions is performed across frames subsequent to the selected frame of the video. Motions of the identified regions are calculated, and new regions for tracking are identified. Note that the term "video" is used herein in its broadest sense to include an electronically stored sequence of still images that represent scenes in motion and "video content" means image information stored in the form of a video.

According to an aspect of the invention, the computer readable program code for causing the computer to of identify new regions for tracking may be responsive to computer readable program code for causing the computer to calculate motions of the identified regions and/or track identified regions across frames. According to another aspect of the invention, the computer readable program code for causing the computer to identify regions of the selected frame may include said computer readable program code for causing the computer to segment at least one frame of the video to partition the frame into coherent regions and/or extract local motion estimations for various points in the selected frame across the frames subsequent to the selected frame.

According to another aspect of the invention, the computer readable program code for causing the computer to track identified regions across the frames subsequent to the selected frame may include computer readable program code for causing the computer to determine a motion of each of the identified regions of the selected frame across the frames subsequent to the selected frame.

According to another aspect of the invention, the computer readable program code for causing the computer to determine a motion of each of the identified regions may include computer readable program code for causing the computer to (i) fit the identified regions to the frames subsequent to the selected frame and/or (ii) estimate local motion within the selected frame. According to another aspect of the invention, the computer readable program code for causing the computer to track identified regions across the frames subsequent to the selected frame may include computer readable program code for causing the computer to (i) determine whether a moveable object is trackable and/or (ii) determine whether one of the frames subsequent to the selected frame is substantially different from the selected frame.

According to other aspects of the invention, the computer readable program code for causing the computer to calculate motions of the identified regions may include computer readable program code for causing the computer to (i) determinie a motion of each of the identified regions of the selected frame across the frames subsequent to the selected frame, (ii) group together ones of the identified regions based on their common motion to generate more complex regions of interest, (iii) determine a camera motion and factoring the camera motion out to identify salient objects that move differently from the camera motion, (iv) identify a salient object and continuing tracking of the salient object, and/or (iv) receive user input designating a salient object of interest.

According to another aspect of the invention, a computer program may include a computer usable medium having computer readable program code embodied therein for generating an indexed video. The computer readable program code may include computer readable program code for causing the computer to identify regions of a frame of the video appropriate for tracking, track the identified regions across subsequent frames of the video, calculate motions of the identified regions, identify coherent video objects, identify new regions for tracking in response to the calculated motions, create tubes that relate to the coherent video objects, and index and store the tubes in a searchable database.

According to another aspect of the invention, tracking may include generating tracking information and tube creation may include collecting the tracking information into space-time threads.

According to another aspect of the invention, tube creation may include identifying common salient objects and joining ones of the tubes of respective ones of the common salient objects.

According to another aspect of the invention, the computer readable program code may further include computer readable program code for causing the computer to analyze tube features to identify the weights and/or, tag and assign weights to each of the tubes. According Io another aspect of the invention, an apparatus for extracting objects from a video includes a local tracking engine configured to identify regions of a selected frame of the video appropriate for tracking. The local tracking engine may include a motion calculation engine. An interframe tracking engine receives the identified regions from the local tracking engine and identifies regions across frames subsequent to the selected frame of the video. The motion calculation engine calculates motions of the identified regions.

According to another aspect of the invention, an apparatus for generating an indexed video includes a frame segmentation engine configured to identify regions of a frame of the video appropriate for tracking; a tracking engine configured receiving and tracking the identified regions across subsequent frames of the video; a motion calculation engine configured to calculate motions of the identified regions and identify coherent video objects; an object recognition engine configured to create tubes that relate to the coherent video objects; and a searchable database configured to index and store the tubes.

Additional objects, advantages and novel features of the invention will be set forth in part in the description which follows, and in part will become apparent to those skilled in the art upon examination of the following and the accompanying drawings or may be learned by practice of the invention. The objects and advantages of the invention may be realized and attained by means of the instrumentalities and combinations particularly pointed out in the appended claims.

Brief Description of the Drawings

The drawing figures depict preferred embodiments of the present invention by way of example, not by way of limitations. In the figures, like reference numerals refer to the same or similar elements.

Figure 1 is a flow diagram of a method according to an embodiment of the invention for extracting objects from a video, generating an indexed video and searching videos based on characteristics of the objects; and

Figure 2 is an apparatus according to an embodiment of the invention for executing software implementing procedures according to aspects and features of the invention; and FIG. 3 is a block diagram of an exemplary computer system for executing a program set of instruction for performing object extraction and/or video indexing.

Detailed Description of a Preferred Embodiment of the Invention

The ensuing description provides exemplary embodiments only, and is not intended to limit the scope, applicability, or configuration of the invention. Rather, the ensuing description of the exemplary embodiments will provide those skilled in the art with an enabling description for implementing an example embodiment of the invention. It should be understood that various changes may be made in the function and arrangement of elements without departing from the spirit and scope of the invention.

Referring to Figure 1 of the drawings, a method according to an embodiment of the invention extracts objects from a video, generates an indexed video and searches videos based on characteristics of the objects. Initially the method creates seeds, i.e., takes a local view of a small number of frames to identify or create "seeds" in the form of salient objects that will be subject to tracking. Thus, at step 101 the next n (e.g., 2 or 3) frames are segmented so as to partition the frames into coherent regions, that is, regions having a common characteristic. For example, regions may be defined based on internally similar photometric characteristics (e.g., color, luminosity, texture, etc.), motion (regions of picture elements that move in a cohesive manner), etc. At step 102 local motions estimations are extracted for various points of a selected (e.g., first) frame across the n frames (i.e., the selected and some number of subsequent frames). Local motion estimations may be performed by techniques and methods such as, but not limited to, the formation of local line averages motion estimations, Lucas-Kanade, etc. At step 103 a determination is made of the motion of each of the regions defined in step 101 of the selected or first frame across the subsequent (e.g., n) frames. This may be performed by calculating motion between the frames by determining an optimal fit over geometric transformations of each of the regions to the next frame alone or in combination with local motion estimates obtained from step 102.

Using motion information from step 103, at step 104 those regions having common motion (e.g., moving in a cohesive manner or having motion consistent with the motion of some larger or more complex grouping of regions) are grouped together. For example, regions having a motion consistent with being part of a larger object may be grouped together to form more complex regions of interest. At step 105 the method corrects for apparent motion caused by changes of viewing position, e.g., camera motion, rather than by displacement of a region under examination. Thus, any camera motion is identified and "factored out" so that only motion of a region relative to a fixed reference point (e.g., a stationary background) is considered. Step 105 thusly identifies salient objects that move differently from the camera motion. In addition, step 105 provides for identification of salient objects as may be manually designated by a user (e.g., user identification by a mouse click of an interesting visual object that should be tracked) and continuation of tracking of salient objects for which tracking has been lost and/or needs adjustment. In the case of the latter two situations the method may tag the designated salient objects for extra scrutiny and/or special processing in an enhanced attempt to initiate and continue tracking. In the case of a lost or troubled track, step 105 may examine the ending point of an object for which tracking was lost in an attempt to reacquire tracking of the object based on new locally identified regions corresponding to the previously tracked object.

Once regions are grouped into salient objects, it may only be desirable or necessary to track some of the constituent regions rather than all those identified as constituting the object. Thus, at step 106, for each salient object, a set of regions is identified including those regions that are optimal for tracking purposes. Step 107 performs a test to determine if all moving objects in a next frame to be examined are trackable. For example, some regions or even entire objects may disappear from a subsequent frame, an object may be reoriented to a degree that constituent regions being tracked are no longer visible, or any other number and type of events may result in a previously tracked object becoming untrackable. In such cases, processing reverts back to step 101 to initiate creation of new seeds, i.e., local identification over some small set of n frames of trackable regions and their corresponding salient objects. Where reseeding is necessary, prior tracking history may be used to optimize identification of regions to be tracked.

If all moving objects remain trackable in a subsequent frame to be considered, then a second test is applied to determine if it is appropriate to, nonetheless, create new seeds. This may be desirable or necessary so as to identify new regions and/or objects that may have been introduced since tracking was last initiated. It may also be desirable to "reseed" the tracking process if the current frame is "stale", i.e., those regions and objects being tracked no longer represent optimal regions to be tracked in view of, for example, object movement, changes in lighting, interaction between and among objects, etc. If the current selection of regions and/or objects is stale (e.g., too old) or if substantial changes affecting tracking have occurred, then processing again reverts back to step 101. Otherwise, moving objects in the next frame are tracked at step 109. If still further frames of the video remain, then processing loops back to step 107 to continue tracking of the previously identified moving objects.

If all frames have been processed (i.e., the test at step 110 is passed), processing continues at step 1 1 1 to collect tracking information into space-time threads or "tubes". The intersection of each tube with a sequence of frames defines areas within each frame containing the selected moving objects previously processed, each tube spanning from the corresponding starting frame for which object tracking began through the ending frame at which tracking was terminated. Associated with each tube is information about the corresponding moving object including that derived from the previously described steps including, by way of example, object luminance and texture, shape, motion information, etc.

At step 1 12 processing is performed to identify common salient objects and thereby join corresponding object-related tubes. That is, tubes that are found to be associated and/or describe the same object may be joined to create one larger tube in time and/or space. This may be necessary, for example, because tracking of an object may have been lost at some point but reacquired in an immediately subsequent or some later frame (i.e., with no or some intervening frames).

Weighting of each tube is performed at step 1 13. Weighting may be, for example, a measure of the importance of a moving object. Various indicia of object importance may be considered including, for example, the amount of time or number of frames an object is tracked relative to total video length and/or in comparison to other tracked objects; the volume or space occupied within the frames by an object; amount, speed and nature of motion of the object; photometric features, object color, etc.

Tube (i.e., object) tagging is accomplished at step 114. That is, the corresponding object may be identified by giving it a name or other descriptive designation and/or associated with other information about the tube. Tagging may be accomplished by manual input by the user and/or by automatic recognition of the corresponding object.

At step 1 15 the tubes and associated object-related data may be indexed and stored in searchable database to complete generation of video indexing. Steps 1 16 through 1 18 represent searching of the database for videos containing targeted objects appearing in the respective videos with a desired importance value. Thus, at step 1 16 a target object or objects are specified together with their associated importance values. The database is then searched at step 1 17 for videos having object-related tubes satisfying or most closely satisfying the search conditions, i.e., having (or not having) the target objects and corresponding importance values. The search concludes at step 1 18 with the identification of videos having or most closely matching the specified criteria.

Although a preferred embodiment of the invention has been described having a specified sequence of steps, other sequences are equally feasible. For example, the motion estimation of special points (e.g., Lucas-Kanade) could precede the segmentation in step 101 of any image and/or the tracking of special points in the image could precede the motion estimation in step 103. While the specified order of steps may be found to be efficient for a particular application and platform, other arrangements of steps may be optimized to other environments. More specifically, the sequence of steps 101 and 102 of Figure 1 may be exchanged with no further changes being required. In addition, functions performed by some steps may be incorporated into other steps. For example, the functionality of step 101 may be embedded into step 104 given that the determination of motion estimation in step 103 would be replaced by the motion estimations provided by step 102.

Figure 2 is a block diagram of an apparatus for extracting object from a video to generating an indexed video. Each of the component parts of the apparatus perform functions corresponding to those set forth in connection with the above described method and may be implemented in various forms including, for example, software modules, firmware, hardware and/or any combination thereof. Thus, a local tracking engine 210 includes software, firmware and devices for performing operations corresponding to steps 101 - 106 with reference to Figure 1 to creates salient object seeds that will be subject to tracking. Frame segmentation engine 211 segments the next n (e.g., 2 or 3) frames to partition the frames into coherent regions. Local estimation engine 212 extracts motion estimations for various points of a first or later selected initial seed frame across the n frames. Motion estimation may take the form of various implementation including, but not limited to, the formation of local line averages motion estimations, Lucas-Kanade, etc. The result of the local motion estimation is then supplied to interframe motion estimation engine 214 of motion calculation engine 213. Interframe motion estimation engine 213 determines the motion of each of the regions defined by frame segmentation engine 21 1 for the selected or first frame across some small number of subsequent (e.g., n) frames. Again, this may be performed by calculating motion between the frames by determining an optimal fit over geometric transformations of each of the regions to the next frame alone or in combination with local motion estimates.

Using motion information from interframe motion estimation engine 214, region grouping engine groups those regions having common motion to generate more complex regions of interest. Relative motion determination engine corrects for apparent motion caused by changes of viewing position and identifies salient objects that move differently from the camera motion. Relative motion determination engine 216 may also provide a user interface (e.g., keyboard, mouse, etc.) to allow a user to manually designate and/or salient objects. Still another functionality that may be included is the capability of relative motion determination engine 216 to continue tracking of salient objects for which tracking has been lost and/or needs adjustment either by automatic recognition or manual operator input.

Once regions are grouped into salient objects by motion calculation engine 213, tracking optimization engine 217 identifies a set of regions for each object that is optimal for tracking purposes. However, if one or more of the objects being tracked is no longer trackable or if the frames used to create the seed are too old (e.g., "stale"), then tracking optimization engine initiates generation of a new seed by engines 21 1 - 216. Otherwise, moving objects in the next frame are tracked by interframe tracking engine 218.

Upon the tracking of all frames of a video by interframe tracking engine 218, tracking information collection engine 219 collects tracking information into space-time threads or "tubes". Object recognition engine 220 operates to identify common salient objects and thereby join corresponding object-related tubes. Weighing engine 221 assigns a weight to each tube based on one or more indicia of object importance such as the amount of time or number of frames an object is tracked, object size and/or color in the frame, etc. Tagging engine 222 tags the tube that is then indexed and stored in searchable database 223 to complete generation of video indexing.

Search engine 224 accepts a user or otherwise designated target object with corresponding importance value and searches searchable database 223 to identify videos containing the targeted objects with the desired importance value. Thus, at step 116 a target object or objects are specified together with their associated importance values. In response to user selection of one or more of the identified videos, the selected video(s) may be retrieved by video output engine 225 and provided to the user by user interface 219 (e.g., a personal computer.)

As before, the present embodiment has been described by way of example including one arrangement of components and sequence of processing. However, other arrangements and sequences are equally feasible. For example, local motion estimation engine 212 might be arranged prior to frame segmentation engine 21 1 with no further changes being required. In addition, functions performed by some engines may be incorporated into others. For example, the functionality performed by frame segmentation engine 211 might be embedded into region grouping engine 215 given that the determination of motion estimation performed by interframe motion estimation engine 214 might be replaced by the motion estimations provided by local motion estimation engine 212.

Figure 3 illustrates an exemplary computer system 300 on which object extraction from video content and the generation of indexed video may be implemented according to one embodiment of the present invention. Central processing unit (CPU) 301 is coupled to system bus 302. CPU 301 may be any general purpose CPU. The present invention is not restricted by the architecture of CPU 301 (or other components of exemplary system 300) as long as CPU 301 (and other components of system 300) supports the inventive operations as described herein. CPU 301 may execute the various logical instructions according to embodiments of the present invention. For example, CPU 301 may execute machine-level instructions according to the exemplary operational flows described above in conjunction with FIGS. 1 and 2.

Computer system 300 also preferably includes random access memory (RAM) 303, which may be SRAM, DRAM, SDRAM, or the like. Computer system 300 preferably includes read-only memory (ROM) 304 which may be PROM, EPROM, EEPROM, or the like. RAM 303 and ROM 304 hold / store user and system data and programs, such as a machine-readable and/or executable program of instructions for object extraction and/or video indexing according to embodiments of the present invention.

Computer system 300 also preferably includes input/output (I/O) adapter 305, communications adapter 311, user interface adapter 308, and display adapter 309. I/O adapter 305, user interface adapter 308, and/or communications adapter 31 1 may, in certain embodiments, enable a user to interact with computer system 300 in order to input information.

I/O adapter 305 preferably connects to storage device(s) 306, such as one or more of hard drive, compact disc (CD) drive, floppy disk drive, tape drive, etc. to computer system 300. The storage devices may be utilized when RAM 303 is insufficient for the memory requirements associated with storing data for operations of the system (e.g., storage of videos and related information). Although RAM 303, ROM 304 and/or storage device(s) 306 may include media suitable for storing a program of instructions for object extraction and/or video indexing according to embodiments of the present invention, those having removable media may also be used to load the program and/or bulk data such as large video files.

Communications adapter 31 1 is preferably adapted to couple computer system 300 to network 312, which may enable information to be input to and/or output from system 300 via such network 312 (e.g., the Internet or other wide-area network, a local-area network, a public or private switched telephony network, a wireless network, any combination of the foregoing). For instance, users identifying or otherwise supplying a video for processing may remotely input access information or video files to system 300 via network 312 from a remote computer. User interface adapter 308 couples user input devices, such as keyboard 313, pointing device 307, and microphone 314 and/or output devices, such as speaker(s) 315 to computer system 300. Display adapter 309 is driven by CPU 301 to control the display on display device 310 to, for example, display information regarding a video being processed and providing for interaction of a local user or system operator during object extraction and/or video indexing operations.

It shall be appreciated that the present invention is not limited to the architecture of system 300. For example, any suitable processor-based device may be utilized for implementing object extraction and video indexing, including without limitation personal computers, laptop computers, computer workstations, and multi-processor servers. Moreover, embodiments of the present invention may be implemented on application specific integrated circuits (ASICs) or very large scale integrated (VLSI) circuits. In fact, persons of ordinary skill in the art may utilize any number of suitable structures capable of executing logical operations according to the embodiments of the present invention. While the foregoing has described what are considered to be the best mode and/or other preferred embodiments of the invention, it is understood that various modifications may be made therein and that the invention may be implemented in various forms and embodiments, and that it may be applied in numerous applications, only some of which have been described herein. It is intended by the following claims to claim any and all modifications and variations that fall within the true scope of the inventive concepts.

It should also be noted and understood that all publications, patents and patent applications mentioned in this specification are indicative of the level of skill in the art to which the invention pertains. All publications, patents and patent applications are herein incoφorated by reference to the same extent as if each individual publication, patent or patent application was specifically and individually indicated to be incorporated by reference in its entirety.

Claims

Claims
1. A method of extracting objects from a video comprising the steps of: identifying regions of a selected frame of the video appropriate for tracking; tracking identified regions across frames subsequent to said selected frame of the video; calculating motions of the identified regions; and identifying new regions for tracking.
2. The method according to claim 1 wherein said step of identifying new regions for tracking is responsive to said step of calculating motions of said identified regions.
3. The method according to claim 1 wherein said step of identifying new regions for tracking is responsive to said step of tracking identified regions across frames.
4. The method according to claim 1 wherein said step of identifying regions of said selected frame includes a step of segmenting at least one frame of the video to partition the frame into coherent regions.
5. The method according to claim 1 wherein said step of identifying regions of said selected frame includes a step of extracting local motion estimations for various points in the selected frame across said frames subsequent to said selected frame .
6. The method according to claim 1 wherein said step of tracking identified regions across said frames subsequent to said selected frame includes a step of determining a motion of each of the identified regions of the selected frame across said frames subsequent to said selected frame.
7. The method according to claim 6 wherein said step of determining a motion of each of the identified regions includes fitting said identified regions to said frames subsequent to said selected frame.
8. The method according to claim 6 wherein said step of determining a motion of each of the identified regions includes a step of estimating local motion within said selected frame.
9. The method according to claim 1 wherein said step of tracking identified regions across said frames subsequent to said selected frame includes a step of determining whether a moveable object is trackable.
10. The method according to claim 1 wherein said step of tracking identified regions across said frames subsequent to said selected frame includes a step of determining whether one of said frames subsequent to said selected frame is substantially different from said selected frame.
1 1. The method according to claim 1 wherein said step of calculating motions of the identified regions includes a step of determining a motion of each of the identified regions of the selected frame across said frames subsequent to said selected frame.
12. The method according to claim 1 wherein said step of calculating motions of the identified regions includes a step of grouping together ones of said identified regions based on their common motion to generate more complex regions of interest.
13. The method according to claim 1 wherein said step of calculating motions of the identified regions includes steps of determining a camera motion and factoring said camera motion out to identify salient objects that move differently from said camera motion.
14. The method according to claim 1 wherein said step of calculating motions of the identified regions includes steps of identifying a salient object and continuing tracking of said salient object.
15. The method according to claim 1 wherein said step of calculating motions of the identified regions includes a step of receiving user input designating a salient object of interest.
16. A method of generating an indexed video comprising the steps of: identifying regions of a frame of the video appropriate for tracking; tracking the identified regions across subsequent frames of the video; calculating motions of the identified regions including identifying coherent video objects; identifying new regions for tracking in response to said calculated motions; creating tubes that relate to said coherent video objects; and indexing and storing said tubes in a searchable database.
17. The method according to claim 16 wherein said step of tracking includes generating tracking information and said step of creating tubes includes a step of collecting said tracking information into space-time threads.
18. The method according to claim 16 wherein said step of creating tubes includes steps of identifying common salient objects and joining ones of said tubes of respective ones of said common salient objects.
19. The method according to claim 16 further comprising steps of tagging and assigning weights to each of said tubes.
20. The method according to claim 19 further comprising a step of analyzing tube features to identify said weights.
21. A computer program comprising: a computer usable medium having computer readable program code embodied therein for extracting objects from a video, the computer readable program code including: computer readable program code for causing the computer to identify regions of a selected frame of the video appropriate for tracking; computer readable program code for causing the computer to track identified regions across frames subsequent to said selected frame of the video; computer readable program code for causing the computer to calculate motions of the identified regions; and computer readable program code for causing the computer to identify new regions for tracking.
22. The computer program according to claim 21 wherein said computer readable program code for causing the computer to identify new regions for tracking is responsive to said computer readable program code for causing the computer to calculate motions of the identified regions.
23. The computer program according to claim 21 wherein said computer readable program code for causing the computer to identify new regions for tracking is responsive to said computer readable program code for causing the computer to track identified regions across said frames subsequent to said selected frame of the video.
24. The computer program according to claim 21 wherein said computer readable program code for causing the computer to identify new regions for tracking includes computer readable program code for causing the computer to segment at least one frame of the video to partition the frame into coherent regions.
25. The computer program according to claim 21 wherein said computer readable program code for causing the computer to identify regions of a selected frame of the video appropriate for tracking includes computer readable program code for causing the computer to extract local motion estimations for various points in the selected frame across said frames subsequent to said selected frame.
26. The computer program according to claim 21 wherein said computer readable program code for causing the computer to track said identified regions across said frames subsequent to said selected frame of the video includes a step of determining a motion of each of the identified regions of the selected frame across said frames subsequent to said selected frame.
27. The computer program according to claim 26 wherein said step of determining a motion of each of the identified regions includes computer readable program code for causing the computer fit said identified regions to said frames subsequent to said selected frame.
28. The computer program according to claim 26 wherein said step of determining a motion of each of the identified regions includes computer readable program code for causing the computer estimate local motion within said selected frame.
29. The computer program according to claim 21 wherein said computer readable program code for causing the computer to track said identified regions across said frames subsequent to said selected frame of the video includes computer readable program code for causing the computer determine whether a moveable object is trackable.
30. The computer program according to claim 21 wherein said computer readable program code for causing the computer to track said identified regions across said frames subsequent to said selected frame of the video includes computer readable program code for causing the computer determine whether one of said frames subsequent to said selected frame is substantially different from said selected frame.
31. The computer program according to claim 21 wherein computer readable program code for causing the computer to calculate motions of the identified regions includes computer readable program code for causing the computer determine a motion of each of the identified regions of the selected frame across said frames subsequent to said selected frame.
32. The computer program according to claim 21 wherein said computer readable program code for causing the computer to calculate motions of the identified regions includes computer readable program code for causing the computer group together ones of said identified regions based on their common motion to generate more complex regions of interest.
33. The computer program according to claim 21 wherein said computer readable program code for causing the computer to calculate motions of the identified regions includes computer readable program code for causing the computer determine a camera motion and factor said camera motion out to identify salient objects that move differently from said camera motion.
34. The computer program according to claim 21 wherein said computer readable program code for causing the computer to calculate motions of the identified regions includes computer readable program code for causing the computer identify a salient object and continue tracking of said salient object.
35. The computer program according to claim 21 wherein said computer readable program code for causing the computer to calculate motions of the identified regions includes computer readable program code for causing the computer to receivine user input designating a salient object of interest.
36. A computer program comprising: a computer usable medium having computer readable program code embodied therein for generating an indexed video, the computer readable program code including: computer readable program code for causing the computer to identify regions of a frame of the video appropriate for tracking; computer readable program code for causing the computer to track the identified regions across subsequent frames of the video; computer readable program code for causing the computer to calculate motions of the identified regions and to identify coherent video objects; computer readable program code for causing the computer to identify new regions for tracking in response to said calculated motions; computer readable program code for causing the computer to create tubes that relate to said coherent video objects; and computer readable program code for causing the computer to index and store said tubes in a searchable database.
37. The computer program according to claim 36 wherein said computer readable program code for causing the computer to track the identified regions across said subsequent frames of the video includes computer readable program code for causing the computer to generate tracking information and said computer readable program code for causing the computer to create said tubes that relate to said coherent video objects includes computer readable program code for causing the computer to collect said tracking information into space-time threads.
38. The computer program according to claim 36 wherein said computer readable program code for causing the computer to create said tubes that relate to said coherent video objects includes computer readable program code for causing the computer to identify common salient objects and join ones of said tubes of respective ones of said common salient objects.
39. The computer program according to claim 36 wherein the computer readable program code further comprises: computer readable program code for causing the computer to tag and assign weights to each of said tubes.
40. The computer program according to claim 39 further wherein the computer readable program code further comprises: computer readable program code for causing the computer to analyze tube features to identify said weights.
41. An apparatus for extracting objects from a video comprising: a local tracking engine configured to identify regions of a selected frame of the video appropriate for tracking, said local tracking engine including a motion calculation engine; and an interframe tracking engine receiving said identified regions from said local tracking engine and configured to identify regions across frames subsequent to said selected frame of the video, said motion calculation engine is configured to calculate motions of the identified regions.
42. The apparatus according to claim 41 wherein said local tracking engine includes a frame segmentation engine configured to segment at least one frame of the video to partition the frame into coherent regions.
43. The apparatus according to claim 41 wherein said local tracking engine includes a local motion estimation engine configured to extract local motion estimations for various points in the selected frame across said frames subsequent to said selected frame.
44. The apparatus according to claim 41 wherein said local tracking engine includes a motion calculation engine configured to determine a motion of each of the identified regions of the selected frame across said frames subsequent to said selected frame.
45. The apparatus according to claim 41 wherein said local tracking engine includes an interframe motion estimation engine configured to determine whether one of said frames subsequent to said selected frame is substantially different from said selected frame.
46. The apparatus according to claim 41 wherein said motion calculation engine includes an interframe calculation engine configured to determine a motion of each of the identified regions of the selected frame across said frames subsequent to said selected frame.
47. The apparatus according to claim 41 wherein said motion calculation engine includes a region grouping engine configured to group together ones of said identified regions based on their common motion to generate more complex regions of interest.
48. The apparatus according to claim 41 wherein said motion calculation engine includes a relative motion determination engine configured to determine a camera motion and factor said camera motion out to identify salient objects that move differently from said camera motion.
49. The apparatus according to claim 41 wherein said motion calculation engine includes a region grouping engine configured to identify a salient object and control tracking of said salient object.
50. The apparatus according to claim 41 wherein said motion calculation engine includes a user input device for receiving user input designating a salient object of interest.
51. An apparatus for generating an indexed video comprising:
a frame segmentation engine configured to identify regions of a frame of the video appropriate for tracking; a tracking engine configured receiving and tracking said identified regions across subsequent frames of the video; a motion calculation engine configured to calculate motions of the identified regions and identify coherent video objects; an object recognition engine configured to create tubes that relate to said coherent video objects; and a searchable database configured to index and store the tubes.
52. The apparatus according to claim 51 wherein said tracking engine is further configured to generate tracking information and said object recognition engine is further configured to collect said tracking information into space-time threads.
53. The apparatus according to claim 51 wherein said object recognition engine is further configured to identify common salient objects and join ones of said tubes of respective ones of said common salient objects.
54. The apparatus according to claim 51 further comprising a weighing engine configured to assign weights to each of said tubes.
55. The apparatus according to claim 54 wherein said weighing engine is further configured to analyze tube features to identify said weights.
PCT/US2007/024197 2006-11-20 2007-11-20 Method of and apparatus for performing motion-based object extraction and tracking in video WO2008063614A2 (en)

Priority Applications (16)

Application Number Priority Date Filing Date Title
US86655206P true 2006-11-20 2006-11-20
US60/866,552 2006-11-20
US86927106P true 2006-12-08 2006-12-08
US86927906P true 2006-12-08 2006-12-08
US60/869,271 2006-12-08
US60/869,279 2006-12-08
US11/687,341 2007-03-16
US11/687,300 2007-03-16
US11/687,290 2007-03-16
US11/687,326 US20080120291A1 (en) 2006-11-20 2007-03-16 Computer Program Implementing A Weight-Based Search
US11/687,326 2007-03-16
US11/687,290 US20080120290A1 (en) 2006-11-20 2007-03-16 Apparatus for Performing a Weight-Based Search
US11/687,341 US8488839B2 (en) 2006-11-20 2007-03-16 Computer program and apparatus for motion-based object extraction and tracking in video
US11/687,300 US20080120328A1 (en) 2006-11-20 2007-03-16 Method of Performing a Weight-Based Search
US11/687,261 2007-03-16
US11/687,261 US8379915B2 (en) 2006-11-20 2007-03-16 Method of performing motion-based object extraction and tracking in video

Publications (2)

Publication Number Publication Date
WO2008063614A2 true WO2008063614A2 (en) 2008-05-29
WO2008063614A3 WO2008063614A3 (en) 2008-08-14

Family

ID=39430363

Family Applications (3)

Application Number Title Priority Date Filing Date
PCT/US2007/024197 WO2008063614A2 (en) 2006-11-20 2007-11-20 Method of and apparatus for performing motion-based object extraction and tracking in video
PCT/US2007/024199 WO2008063616A2 (en) 2006-11-20 2007-11-20 Apparatus for and method of robust motion estimation using line averages
PCT/US2007/024198 WO2008063615A2 (en) 2006-11-20 2007-11-20 Apparatus for and method of performing a weight-based search

Family Applications After (2)

Application Number Title Priority Date Filing Date
PCT/US2007/024199 WO2008063616A2 (en) 2006-11-20 2007-11-20 Apparatus for and method of robust motion estimation using line averages
PCT/US2007/024198 WO2008063615A2 (en) 2006-11-20 2007-11-20 Apparatus for and method of performing a weight-based search

Country Status (1)

Country Link
WO (3) WO2008063614A2 (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102073668A (en) * 2009-11-23 2011-05-25 英特尔公司 Searching and extracting digital images from digital video files
EP3264394A1 (en) * 2016-06-30 2018-01-03 LACS S.r.l. A method and a system for monitoring military tactics simulations

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2011233039A (en) * 2010-04-28 2011-11-17 Sony Corp Image processor, image processing method, imaging device, and program
KR20160075126A (en) * 2014-12-19 2016-06-29 삼성전자주식회사 Method of providing content and electronic apparatus thereof

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020114394A1 (en) * 2000-12-06 2002-08-22 Kai-Kuang Ma System and method for motion vector generation and analysis of digital video clips
US6643387B1 (en) * 1999-01-28 2003-11-04 Sarnoff Corporation Apparatus and method for context-based indexing and retrieval of image sequences
US6766037B1 (en) * 1998-10-02 2004-07-20 Canon Kabushiki Kaisha Segmenting moving objects and determining their motion

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5886745A (en) * 1994-12-09 1999-03-23 Matsushita Electric Industrial Co., Ltd. Progressive scanning conversion apparatus
US5745890A (en) * 1996-08-09 1998-04-28 Digital Equipment Corporation Sequential searching of a database index using constraints on word-location pairs
US6829604B1 (en) * 1999-10-19 2004-12-07 Eclipsys Corporation Rules analyzer system and method for evaluating and ranking exact and probabilistic search rules in an enterprise database
JP4612760B2 (en) * 2000-04-25 2011-01-12 キヤノン株式会社 Image processing apparatus and method
US7003513B2 (en) * 2000-07-04 2006-02-21 International Business Machines Corporation Method and system of weighted context feedback for result improvement in information retrieval
KR100468171B1 (en) * 2002-12-13 2005-01-26 삼성전자주식회사 System and method for searching information of broadcasting program
US7440964B2 (en) * 2003-08-29 2008-10-21 Vortaloptics, Inc. Method, device and software for querying and presenting search results

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6766037B1 (en) * 1998-10-02 2004-07-20 Canon Kabushiki Kaisha Segmenting moving objects and determining their motion
US6643387B1 (en) * 1999-01-28 2003-11-04 Sarnoff Corporation Apparatus and method for context-based indexing and retrieval of image sequences
US20020114394A1 (en) * 2000-12-06 2002-08-22 Kai-Kuang Ma System and method for motion vector generation and analysis of digital video clips

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102073668A (en) * 2009-11-23 2011-05-25 英特尔公司 Searching and extracting digital images from digital video files
EP3264394A1 (en) * 2016-06-30 2018-01-03 LACS S.r.l. A method and a system for monitoring military tactics simulations

Also Published As

Publication number Publication date
WO2008063614A3 (en) 2008-08-14
WO2008063616A2 (en) 2008-05-29
WO2008063616A3 (en) 2008-08-07
WO2008063615A2 (en) 2008-05-29
WO2008063615A3 (en) 2008-10-30

Similar Documents

Publication Publication Date Title
Rui et al. Constructing table-of-content for videos
US7024020B2 (en) Apparatus and method for generating object-labeled image in video sequence
Stringa et al. Real-time video-shot detection for scene surveillance applications
Flickner et al. Query by image and video content: The QBIC system
DE69723346T2 (en) Method for retrieving images based on a probability function
Hampapur et al. Production model based digital video segmentation
US6331859B1 (en) Video skimming system utilizing the vector rank filter
Föckler et al. PhoneGuide: museum guidance supported by on-device object recognition on mobile phones
JP4990383B2 (en) Image group expression method, image group search method, apparatus, computer-readable storage medium, and computer system
JP4702877B2 (en) Display device
Ngo et al. On clustering and retrieval of video shots through temporal slices analysis
US5708767A (en) Method and apparatus for video browsing based on content and structure
Cong et al. Towards scalable summarization of consumer videos via sparse dictionary selection
US8379987B2 (en) Method, apparatus and computer program product for providing hand segmentation for gesture analysis
JP5318115B2 (en) Image classification by location
US8170280B2 (en) Integrated systems and methods for video-based object modeling, recognition, and tracking
Smeaton et al. High-level feature detection from video in TRECVid: a 5-year retrospective of achievements
Torralba et al. Labelme: Online image annotation and applications
US6690725B1 (en) Method and a system for generating summarized video
Cernekova et al. Information theory-based shot cut/fade detection and video summarization
Greenspan et al. Probabilistic space-time video modeling via piecewise GMM
KR101640268B1 (en) Method and system for automated annotation of persons in video content
Isola et al. What makes an image memorable?
US8917902B2 (en) Image overlaying and comparison for inventory display auditing
Ardizzone et al. Automatic video database indexing and retrieval

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 07862127

Country of ref document: EP

Kind code of ref document: A2

NENP Non-entry into the national phase in:

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC, EPO FORM 1205A SENT ON 13/10/09

122 Ep: pct application non-entry in european phase

Ref document number: 07862127

Country of ref document: EP

Kind code of ref document: A2