System and method for quick object verification

• US 8,854,474 B2
• Filed: 03/08/2011
• Issued: 10/07/2014
• Est. Priority Date: 03/08/2011
• Status: Active Grant

• Current Assignee: Qognify Ltd. (PSBU Midco SARL)
• Original Assignee: Nice Systems Limited (Nice Ltd)
• Inventors: Blumstein-Koren, Guy, Yehezkel, Raanan Yonatan
• Primary Examiner(s): Perungavoor, Sath V
• Assistant Examiner(s): WILLIAMS, JEFFERY A
• Application Number:
  US13/042,483
• Publication Number:
  US 20120229629A1
• Time in Patent Office: 1,309 Days
• Field of Search: 348/169, 348/208.14, 382/103
• US Class Current: 348/169
• CPC Class Codes: G06F 16/7837 using objects detected or recognised in the video content

Abstract

Systems and methods for identifying an appearance of an object in a video stream, include receiving, in real-time, an indication of an area of interest, from a user, within a frame of a video stream. The area of interest may include an image of an object. Automatically analyzing the characteristics of the area of interest, finding a frame in which the object first appears in time in the video stream and displaying to the user the video stream starting from the frame in which the object first appears.

Claims

1. A method for real-time investigation of objects detected in an area monitored by a closed circuit television (CCTV) system, the method comprising:

receiving a plurality of video streams of the monitored area from one or more cameras;

receiving by a computing device a query initiated by a user of the CCTV system upon becoming aware that a suspicious object is present in the monitored area, wherein the query includes an indication of an area of interest, marked by the user within a first frame of a video stream capturing the monitored area, wherein the area of interest includes an image of the suspicious object and the query is to scan the video streams backwards and identify a first appearance of the suspicious object in the video streams;

automatically scanning one or more relevant video streams of the plurality of video streams backwards in time and analyzing the area of interest in previous frames;

automatically identifying the first appearance of the suspicious object in a second frame, the second frame being previous in time to the first frame and wherein the second frame is from the one or more relevant video streams;

and

displaying to the user a playback of the video stream starting from the second frame or from a frame previous in time to the second frame so as to enable the real-time investigation of the suspicious object.

2. The method of claim 1, wherein initiating the query is based on a notification regarding the suspicious object received from a person present at the monitored area and the area of interest is determined based on the notification.

3. The method of claim 1, further comprising displaying the video stream backward starting from the second frame.

4. The method of claim 3, further comprising:

receiving by the computing device an additional query initiated by the user of the CCTV system upon investigating the playback of the video stream, the additional query requesting searching backwards starting from a third frame that is previous in time to the second frame;

automatically scanning one or more relevant video streams backward in time from the third frame and analyzing the area of interest in previous frames until another first appearance is found.

5. The method of claim 1, wherein automatically identifying the first appearance comprises:

extracting a feature vector from the area of interest of a plurality of frames of the video stream;

comparing the query feature vector to the feature vector of each of the plurality of frames;

determining whether a mismatch exists between the first frame and each one of the plurality of frames and counting number of mismatches;

comparing the number of mismatches to a predefined threshold;

if the number of mismatches is higher than the predefined threshold marking a second frame as the first appearance of the object in the scene;

if the number of mismatches is lower than the predefined threshold comparing the query feature vector to the feature vector of a next frame.

6. A video analytics system comprising:

a plurality of cameras of a closed circuit television (CCTV) system;

a data server to save a video stream of a monitored area recorded by the plurality of cameras;

a computing device to;

receive a plurality of video streams of the monitored area from the plurality of cameras;

receive a query initiated by a user of the CCTV system upon becoming aware that a suspicious object is present in the monitored area, wherein the query includes an indication of an area of interest marked by the user within a first frame of the video stream, wherein the area of interest includes an image of the suspicious object and the query is to scan the video streams backwards and identify a first appearance of the suspicious object in the video streams;

automatically scan one or more relevant video streams of the plurality of video streams backwards in time and analyzing the area of interest in previous frames;

automatically identify the first appearance of the suspicious object in a second frame, the second frame being previous in time to the first frame and wherein the second frame is from the one or more relevant video streams;

and

an output device to display to the user a playback of the video stream starting from the second frame or from a frame previous in time to the second frame so as to enable the real-time investigation of the suspicious object.

7. The system of claim 6, further comprising:

a communication element to transfer a notification regarding the suspicious object from a person present at the monitored area, wherein the area of interest is based on the notification.

8. The system of claim 6, wherein the computing device is further to extract a query feature vector from the area of interest of the first frame, extract a feature vector from the area of interest of a plurality of frames of the video stream to compare the query feature vector to the feature vector of each of the plurality of frames and to determine whether a mismatch between the first frame and each one of the plurality of frames exists.

9. The system of claim 6, wherein the computing device is to receive the area of interest from an input device.

10. The system of claim 6, wherein the data server is to compress the video stream and transfer a compressed video stream to the computing device.

11. The system of claim 6, wherein the computing device is to decompress a selected subset of the plurality of frames of the video stream.

12. An article comprising a non-transitory computer-readable storage medium, having stored thereon instructions, that when executed on a computer, cause the computer to:

receive a plurality of video streams of a monitored area from one or more cameras;

receive a query initiated by a user of a closed circuit television (CCTV) system upon becoming aware that a suspicious object is present in the monitored area, wherein the query includes an indication of an area of interest, marked by the user within a first frame of a video stream recorded at a capturing the monitored area, wherein the area of interest includes an image of the suspicious object and the query is to scan the video streams backwards and identify a first appearance of the suspicious object in the video streams;

automatically scan one or more relevant video streams of the plurality of video streams backwards in time and analyzing the area of interest in previous frames;

automatically identify the first appearance of the suspicious object in a second frame, the second frame being previous in time to the first frame and wherein the second frame is from the one or more relevant video streams; and

display to the user a playback of the video stream starting from the second frame or from a frame previous in time to the second frame so as to enable the real-time investigation of the suspicious object.