Categorizing moving objects into familiar colors in video

• US 8,223,835 B2
• Filed: 03/20/2007
• Issued: 07/17/2012
• Est. Priority Date: 03/20/2007
• Status: Expired due to Fees

• Current Assignee: Kyndryl Incorporated
• Original Assignee: International Business Machines Corporation
• Inventors: Brown, Lisa M.
• Primary Examiner(s): Shiferaw, Eleni
• Assistant Examiner(s): Teslovich, Tamara
• Application Number:
  US11/688,588
• Publication Number:
  US 20080232685A1
• Time in Patent Office: 1,946 Days
• Field of Search: 375/240.03
• US Class Current: 375/240.03
• CPC Class Codes: G06T 7/20 Analysis of motion motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51
• CPC Class Codes: G06V 10/56 relating to colour

Abstract

An improved solution for categorizing moving objects into familiar colors in video is provided. In an embodiment of the invention, a method for categorizing moving objects into familiar colors in video comprises: receiving a video input; determining at least one object track of the video input; creating a normalized cumulative histogram of the at least one object track; and one of: performing a parameterization quantization of the histogram including separating the histogram into regions based on at least one surface curve derived from one of saturation and intensity; or identifying a significant color of the quantized histogram.

Claims

1. A method for categorizing moving objects into familiar colors in video comprising:

receiving a video input;

automatically determining, using a computer device, at least one object track of the video input, each of the at least one object track tracking a movement of an object within the video input;

creating a normalized cumulative histogram of the at least one object track;

one of;

performing a parameterization quantization of the histogram including separating the histogram into regions based on at least one surface curve derived from one of saturation and intensity;

or

identifying a significant color of the histogram; and

sub sampling a plurality of frames of the at least one object track according to the equation i>

S+M*2ⁿ, where i is a current number of frames in the object track, S is a constant based on a number of frames needed before sampling starts, M is an initial number of frames that can be skipped, and n is an iterator.

2. The method of claim 1, further comprising outputting one of the parameterization quantization and the significant color.

3. The method of claim 1, the performing comprising:

separating a color space into chromatic and achromatic regions; and

dividing the regions using at least one threshold.

4. The method of claim 1, the performing comprising setting quantization parameters that include one of a feedback visualization of discretization of a color or color classification results of moving objects from a database.

5. The method of claim 1, further comprising using an output of the histogram to derive one of quantization and significant color identification parameters.

6. The method of claim 1, further comprising converting the at least one object track to a hue/saturation/intensity (HSI) space.

7. The method of claim 1, the identifying comprising selecting hue over black and white using a per color threshold.

8. The method of claim 1, the determining comprising:

outputting a segmented foreground region; and

associating the segmented foreground region to the at least one object track.

9. A system for categorizing moving objects into familiar colors in video, the system comprising:

a system for receiving a video input;

a system for automatically determining at least one object track of the video input, each of the at least one object track tracking a movement of an object within the video input;

a system for creating a normalized cumulative histogram of the at least one object track;

one of;

a system for performing a parameterization quantization of the histogram including separating the histogram into regions based on at least one surface curve derived from one of saturation and intensity;

or

a system for identifying a significant color of the histogram; and

a system for sub sampling a plurality of frames of the at least one object track according to the equation i>

S+M*2ⁿ, where i is a current number of frames in the object track, S is a constant based on a number of frames needed before sampling starts, M is an initial number of frames that can be skipped, and n is an iterator.

10. The system of claim 9, further comprising a system for outputting one of the parameterization quantization and the significant color.

11. The system of claim 9, the system for performing comprising:

a system for separating a color space into chromatic and achromatic regions; and

a system for dividing the regions using at least one threshold.

12. The system of claim 9, the system for performing comprising a system for setting quantization parameters that include one of a feedback visualization of discretization of a color or color classification results of moving objects from a database.

13. The system of claim 9, further comprising a system for using an output of the histogram to derive one of quantization and significant color identification parameters.

14. The system of claim 9, further comprising a system for converting the at least one object track to a hue/saturation/intensity (HSI) space.

15. The system of claim 9, the system for identifying comprising a system for selecting hue over black and white using a per color threshold.

16. The system of claim 9, the system for determining comprising:

a system for outputting a segmented foreground region; and

a system for associating the segmented foreground region to the at least one object track.

17. A computer program comprising program code stored on a computer-readable device, which when executed, enables a computer system to implement a method of categorizing moving objects into familiar colors in video, the method comprising:

receiving a video input;

automatically determining at least one object track of the video input, each of the at least one object track tracking a movement of an object within the video input;

creating a normalized cumulative histogram of the at least one object track;

one of;

performing a parameterization quantization of the histogram including separating the histogram into regions based on at least one surface curve derived from one of saturation and intensity;

or

identifying a significant color of the histogram; and

sub sampling a plurality of frames of the at least one object track according to the equation i>

S+M*2ⁿ, where i is a current number of frames in the object track, S is a constant based on a number of frames needed before sampling starts, M is an initial number of frames that can be skipped, and n is an iterator.

18. A method of generating a system for categorizing moving objects into familiar colors, the method comprising:

providing a computer system operable to;

receive a video input;

automatically determine at least one object track of the video input, each of the at least one object track tracking a movement of an object within the video input;

create a normalized cumulative histogram of the at least one object track;

one of;

perform a parameterization quantization of the histogram including separating the histogram into regions based on at least one surface curve derived from one of saturation and intensity;

or

identify a significant color of the histogram; and

sub sample a plurality of frames of the at least one object track according to the equation i>

S+M*2ⁿ, where i is a current number of frames in the object track, S is a constant based on a number of frames needed before sampling starts, M is an initial number of frames that can be skipped, and n is an iterator.