Method and system for rapid matching of video streams

• US 8,031,947 B2
• Filed: 03/24/2008
• Issued: 10/04/2011
• Est. Priority Date: 04/03/2007
• Status: Expired due to Fees

• Current Assignee: Kenneth P. Jacobsen
• Original Assignee: Kenneth P. Jacobsen
• Inventors: Jacobsen, Kenneth P.
• Primary Examiner(s): WERNER, BRIAN P
• Application Number:
  US12/079,006
• Publication Number:
  US 20080247663A1
• Time in Patent Office: 1,289 Days
• Field of Search: None
• US Class Current: 382/199
• CPC Class Codes: G06F 16/7854 using shape G06F16/7837 takes precedence

Abstract

Techniques for managing video stream data and portions of video stream data are disclosed. In particular, the present invention enables the user to match portions of video streams against stored, previously characterized video streams using Canny edge detection filtering coupled with Hu invariant third moments of those edges. According to one aspect of the present invention, the video stream data is encoded with the following technique, which selects specific frames from the video stream, finds the edges of objects within these selected frames using Canny edge detection, separates each edge into a distinct object, calculates Hu invariant third moments for each edge, and stores each edge together with the video stream and frame identification and it'"'"'s Hu invariant third moments in a database for later comparison. Similar encoding can then be performed for a query video stream to compare against the previously identified video streams stored in the database to identify a unique video stream. It is emphasized that the present invention applies to video stream data but can also be used for imagery data in any other form or taken in any spectrum.

Claims

1. A method of characterizing digital video data with arbitrary image edges contained therein, the characterized video data is then stored in an image edge database for rapid matching of video streams, said method comprises:

receiving a digital video data stream;

segmenting the received video data stream into at least one frame of pixel information;

detecting and determining one or more image edges using an edge filter applied to the pixel information;

calculating Hu invariant third moments for each of the detected image edges, wherein said each of the detected image edges is assigned a unique identifier based on the Hu invariant third moments;

converting the calculated Hu invariant third moments into a set of integer indices based on a set of predefined criteria; and

storing the unique identifier of said each of the detected image edges into an image edge database with a database location in reference to the set of integer indices.

2. The method of claim 1, further comprises creating at least one intensity balanced image from each of the at least one frame.

3. The method of claim 2, wherein each of the at least one intensity balanced image is created from a gray-scale composite of multiple color planes of the pixel information.

4. The method of claim 2, wherein each of the at least one intensity balanced image is created from individual color planes of the pixel information.

5. The method of claim 2, wherein each of the at least one intensity balanced image is created from spectral and composite image representations of the pixel information.

6. The method of claim 1, wherein the edge filter includes, but is not limited to, Canny edge filter, Sobel edge filter, gradient edge filter, Laplacian edge filter, Prewitt edge filter, Prewitt-Canny edge filter, Susan edge filter, Roberts edge filter, and Marr-Hildreth edge filter.

7. The method of claim 1, wherein the Hu invariant third moments comprise seven terms that are invariant with respect to translation, rotation and scaling of said each of the detected image edges within the frame of pixel information.

8. The method of claim 7, wherein the set of integer indices contains at least one dimension, each corresponding to one of the seven terms of the Hu invariant third moments.

9. The method of claim 8, wherein each of the set of integer indices represents a specific range in the corresponding one of the seven terms of the Hu invariant third moments.

10. A system for characterizing digital video data with arbitrary image edges contained therein, the characterized video data is then stored in an image edge database for rapid matching of video streams, said system comprises:

a main memory for storing computer readable code for an application module; and

at least one processor coupled to the main memory, said at least one processor executing the computer readable code in the main memory to cause the application module to perform operations of;

receiving a video data stream;

segmenting the received video stream into at least one frame of pixel information;

detecting and determining one or more image edges using an edge filter applied to the pixel information;

calculating Hu invariant third moments for each of the detected image edges, wherein said each of the detected image edges is assigned a unique identifier based on the Hu invariant third moments;

converting the calculated Hu invariant third moments into a set of integer indices based on a set of predefined criteria; and

storing the unique identifier of said each of the detected image edges into an image edge database with a database location in reference to the set of integer indices.

11. The system of claim 10, further comprises creating at least one intensity balanced image from each of the at least one frame.

12. The system of claim 10, wherein the Hu invariant third moments comprise seven terms that are invariant with respect to translation, rotation and scaling of said each of the detected image edges within the frame of pixel information.

13. The system of claim 12, wherein the set of integer indices contains at least one dimension, each corresponding to one of the seven terms of the Hu invariant third moments.

14. The system of claim 13, wherein each of the set of integer indices represents a specific range in the corresponding one of the seven terms of the Hu invariant third moments.

15. A non-transitory computer usable medium having computer executable instructions stored thereon for controlling a computer system for characterizing digital video data with arbitrary image edges contained therein, the characterized video data is then stored in an image edge database for rapid matching of video streams by a method comprising:

receiving a video data stream;

segmenting the received video stream into at least one frame of pixel information;

detecting and determining one or more image edges using an edge filter applied to the pixel information;

calculating Hu invariant third moments for each of the detected image edges, wherein said each of the detected image edges is assigned a unique identifier based on the Hu invariant third moments;

converting the calculated Hu invariant third moments into a set of integer indices based on a set of predefined criteria; and

storing the unique identifier of said each of the detected image edges into an image edge database with a database location in reference to the set of integer indices.

16. The non-transistory computer usable medium of claim 15, wherein said method further comprises creating at least one intensity balanced image from each of the at least one frame.

17. The non-transitory computer usable medium of claim 15, wherein the Hu invariant third moments comprise seven terms that are invariant with respect to translation, rotation and scaling of said each of the detected image edges within the frame of pixel information.

18. The non-transitory computer usable medium of claim 17, wherein the set of integer indices contains at least one dimension, each corresponds to one of the seven terms of the Hu invariant third moments.

19. The non-transitory computer usable medium of claim 18, wherein each of the set of integer indices represents a specific range in the corresponding one of the seven terms of the Hu invariant third moments.

20. The non-transitory computer usable medium of claim 17, wherein the image edge database comprises a plurality of bins, each of the bins corresponds to one of the set of integer indices.