Method for dominant color setting of video region and data structure and method of confidence measure extraction

• US 20050220339A1
• Filed: 05/20/2005
• Published: 10/06/2005
• Est. Priority Date: 01/29/1999
• Status: Active Grant

• Current Assignee: LG Electronics, Inc. (LG Corporation)
• Original Assignee: LG Electronics, Inc. (LG Corporation)
• Inventors: Kim, Hyeon Jun
• Granted Patent: US 7,760,935 B2
• Time in Patent Office: Days
• US Class Current: 382/165
• CPC Class Codes: G06F 16/785 using colour or luminescence
• CPC Class Codes: G06V 10/56 relating to colour
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A method for a dominant color setting of a video region and a data structure and a method of a confidence measure extraction are disclosed. The video region dominant color setting method is characterized in that a region dominant color descriptor is expressed by the number of dominant colors with respect to a certain region, a dominant color expressed, a frequency that the dominant color appears, and an accuracy of a color value representing the region in a region dominant color based on various region dominant color extraction methods, for thereby expressing a region dominant color using a plurality of colors with respect to a region dominant color value and a confidence value of a region dominant color information based on various region dominant color feature extracting methods.

Claims

1. A dominant color setting method which targets a content-based retrieval for color in visual data, comprising:

generating a region dominant color descriptor incorporating information indicating a number of dominant colors with respect to a region of interest in visual data, at least one expressed dominant color, a frequency with which the dominant color appears in the region, and an accuracy of a color value representing the region.

2. The method of claim 1, wherein the region dominant color descriptor is a first dominant color descriptor formed based on one dominate color extraction method, and wherein said method further comprises:

forming a second dominant color descriptor by transforming data formed by another dominant color descriptor extraction method into a certain inherent data format in a different system based on the formalized description with respect to the extraction method of the region dominant color descriptor; and

comparing the first dominant color descriptor to the second dominant color descriptor.

3. The method of claim 1, wherein an expression accuracy of the dominant colors extracted by a certain method is obtained in accordance with a degree of confidence of the region dominant color descriptor.

4. The method of claim 1, wherein said accuracy information includes or corresponds to confidence information determined based on a maximum color variation value in which the color which is recognized in accordance with an increase/decrease of the color expressed by a certain value is the same color.

5. The method of claim 1, wherein when a certain color is mapped as a dominant color in an image region, color variance which is a difference between an accurate value of the color and the dominant color value is adapted to confidence information included in or corresponding to said accuracy information.

6. The method of claim 1, wherein a coherency value which represents a concentration degree of the pixels of a color with respect to the dominant color is adapted to confidence information included in or corresponding to said accuracy information.

7. The method of claim 1, wherein a size of the region that a dominant color covers in the image region is adapted to confidence information included in or corresponding to said accuracy information.

8. The method of claim 1, wherein a position of each color pixel in the image region is adapted to confidence included in or corresponding to said accuracy information.

9. The method of claim 1, wherein said accuracy information includes or corresponds to a confidence measure expressed by a vector value based on a normalized coherence average value, an average value with respect to a difference when a certain color is recognized as a dominant color, a value obtained by summing the size that the dominant color covers in all image regions, and an average value with respect to a position in a region of each color pixel.

10. The method of claim 1, wherein an interoperability between different feature extraction methods is implemented by comparing at least one confidence value expressed based on the region dominant color descriptor obtained by different region dominant color feature extraction methods with a region dominant color value.

11. The method of claim 1, wherein said accuracy information includes or corresponds to a confidence measure expressed with respect to each dominant color of the region dominant color descriptor.

12. The method of claim 1, wherein the accuracy information includes or corresponds to a confidence measure per each dominant color expressed by a vector that includes elements or a subset of the elements of spatial variance, color variance, the size of a region that the dominant color covers and a position of dominant color pixels in the region.

13. An interoperability maintaining method between different retrieval systems, comprising:

transforming, comparing and searching a sharing data format using a region descriptor of each system, a region dominant color descriptor of each system, and a region dominant color descriptor extraction method description data of each system.

14. A method for describing dominant color of visual data, comprising:

selecting a region of interest from a media object; and

generating a dominant color descriptor for the region, said descriptor including;

a) information indicative of a number of colors (N) selected for extraction from the region, where N≧

1;

b) information indicative of color values determined for respective ones of the N colors;

c) information indicative of frequencies with which respective ones of the N colors appear in the region; and

d) information indicative of an accuracy of a representative color value for the region, said representative color value determined based on the information in at least one of b) and c).

15. The method of claim 14, wherein the color values in b) are determined based on any one of the following extraction methods:

an average-color method;

a method of expressing only one most frequency appearing color in the region;

a method of expressing more than one most frequently appearing color in the region;

a method of determining which colors appear in the region more than a predetermined percentage of a threshold value; and

a histogram method.

16. The method of claim 14, wherein the information in c) is determined based on pixel counts for respective ones of the N colors.

17. The method of claim 14, wherein the representative color value corresponds to color value in b) when N=1.

18. The method of claim 14, wherein the color values in b) are defined by at least one of color space information, quantization information, color clustering information, and channel information.

19. The method of claim 14, wherein the information in d) is computed by:

a confidence measure which is determined by a vector value based on a normalized coherency value, an average value with respect to a difference when a certain color is recognized as a dominant color, a value obtained by summing the size that the dominant color covers in all image regions, and an average value in a region of each color pixel.

20. A method for describing dominant color in visual data, comprising:

selecting a region of interest from a media object; and

generating a dominant color descriptor for the region, said descriptor including;

a) information indicative of at least one color selected for the region, and

b) information indicative of accuracy of a color value assigned to the region, said color value based on the information in (a).

21. A computer-readable medium for setting color information for visual data, having stored thereon:

a) information indicative of a number of colors (N) selected for extraction from a region of a media object, where N≧

1;

b) information indicative of color values determined for respective ones of the N colors;

c) information indicative of frequencies with which respective ones of the N colors appear in the region; and

d) information indicative of an accuracy of a representative color value for the region, said representative color value determined based on the information in at least one of b) and c).

22. The medium of claim 21, wherein the representative color value corresponds to color value in b) when N=1.

23. The medium of claim 21, wherein the color values in b) are defined by at least one of color space information, quantization information, color clustering information, and channel information.

24. The medium of claim 21, wherein the information in d) is computed by:

a confidence measure which is determined by a vector value based on a normalized coherency value, an average value with respect to a difference when a certain color is recognized as a dominant color, a value obtained by summing the size that the dominant color covers in all image regions, and an average value in a region of each color pixel.