ADAPTIVE HISTOGRAM EQUALIZATION FOR IMAGES WITH STRONG LOCAL CONTRAST

Abstract

The present invention relates to a method, apparatus and computer program product for contrast enhancement of images based on adaptive histogram equalization. In particular it relates to preventing adaptive histogram equalization from causing fading artifacts and object extension artifacts. An adaptive histogram equalization method is provided comprising the steps of dividing an image into regions of pixels, determining structures of local pixel value differences of a predefined strength of the image, building for every region a histogram of the pixel values based on the determined structures of local pixel value differences and mapping pixel values of each region based on the histogram corresponding to the region.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 shows a block diagram of an adaptive histogram equalization method. Optional steps of the method are included.

FIG. 2 shows a block diagram of an adaptive histogram equalization processor. Optional devices of the processor are included.

FIG. 3 shows a flow diagram describing aspects of an embodiment of the adaptive histogram equalization method.

FIG. 4 shows the horizontal overlapping of the block structure used in the embodiment.

FIG. 5 shows an original image.

FIG. 6 shows the structures of local pixel value of the original image.

FIG. 7 shows two difference pictures between processed and original image.

Claims

1. An adaptive histogram equalization method for enhancing the contrast of digital images comprising the steps of dividing an image into regions of pixels,determining structures of local pixel value differences of a predefined strength of the image,building for every region a histogram of the pixel values based on the determined structures of local pixel value differences andmapping pixel values of each region based on the histogram corresponding to the region.

2. An adaptive histogram equalization method according to claim 1, wherein for every region with a determined structure of local pixel value difference a first side and a second side opposite to said first side in relation to the determined structure are defined.

3. An adaptive histogram equalization method according to claim 2, wherein for regions with a determined structure of local pixel value differences the histogram is built from the pixel values of the pixels located on said first side of the determined structure.

4. An adaptive histogram equalization method according to claim 2, wherein for regions with a determined structure of local pixel value differences the amount of pixels located on said second side of the structure modifies the histogram, whereby the larger the number of pixels on said second side the higher the degree of balance or of uniform distribution of the histogram obtained in said building step.

5. An adaptive histogram equalization method according to claim 4, wherein the pixels on said second side are accounted for in the histogram as if one part of the pixels were carrying a minimum value and the other part a maximum value.

6. An adaptive histogram equalization method according to claim 2, further comprising a step of smoothing the histograms or smoothing functions derived from the histograms and used, directly or indirectly, in the step of mapping pixel values.

7. An adaptive histogram equalization method according to claim 2, wherein for regions with no determined structure of local pixel value differences the histogram is built from the pixel values of all pixels of the region.

8. An adaptive histogram equalization method according to claim 2, wherein said structures of local pixel value differences of a predefined strength are determined by an edge detection method, for example the Sobel operator.

9. An adaptive histogram equalization method according to claim 8, wherein single and sparsely distributed edge points determined by the edge detection method are deleted and/or broken edges are connected by means of e.g. morphological filters.

10. An adaptive histogram equalization method according to claim 2, further comprising a step of determining skin colored pixels, whereby for regions with skin colored pixels the amount of skin colored pixels modifies the histogram, whereby the larger the number of skin colored pixels in the region the higher the degree of balance or of uniform distribution of the histogram obtained in said building step.

11. An adaptive histogram equalization method according to claim 2, wherein the histograms are homogenized.

12. An adaptive histogram equalization method according to claim 2, wherein the regions are overlapping and wherein the values of pixels located in the overlapping part of a region and the values of pixels located in the non-overlapping part of a region are mapped only once.

13. An adaptive histogram equalization method according to claim 2, wherein the pixel values are values relating to the brightness and/or colorness of the pixels, for example the luminance value and/or chrominance or chrominance difference values of a YUV, YIQ, or YCbCr or other color model based image format.

14. An adaptive histogram equalization processor for enhancing the contrast of digital images comprising a region specifying device for dividing an image into regions of pixels,a structure determining device for determining structures of local pixel value differences of a predefined strength of the image,a histogram building device for building for every region a histogram of the pixel values, based on the determined structures of local pixel value differences and a mapping device for mapping pixel values of each region based on the histogram corresponding to the region.

15. An adaptive histogram equalization processor according to claim 14, wherein for every region with a determined structure of local pixel value difference a first side and a second side opposite to said first side in relation to the determined structure are defined.

16. An adaptive histogram equalization processor according to claim 15, wherein for regions with a determined structure of local pixel value differences the histogram building device is adapted to build the histogram from the pixel values of the pixels located on said first side of the determined structure.

17. An adaptive histogram equalization processor according to claim 15, wherein for regions with a determined structure of local pixel value differences the histogram building device is adapted to modify the histogram based on the amount of pixels located on said second side of the determined structure, whereby the larger the number of pixels on said second side the higher the degree of balance or of uniform distribution of the histogram obtained by the histogram building device.

18. An adaptive histogram equalization processor according to claim 17, wherein the histogram building device is adapted to account for the pixels on said second side in the histogram as if one part of the pixels were carrying a minimum value and the other part a maximum value.

19. An adaptive histogram equalization processor according to claim 14, further comprising a filtering device for smoothing the histograms or smoothing functions derived from the histograms and used, directly or indirectly, by the mapping device for the mapping of the pixel values.

20. An adaptive histogram equalization processor according to claim 14, wherein for regions with no determined structure of local pixel value differences the histogram building device is adapted to build the histogram from the pixel values of all pixels of the region.

21. An adaptive histogram equalization processor according to claim 14, wherein the structure determining means is adapted to detect said structures of local pixel value differences of a predefined strength by an edge detection method, for example the Sobel operator.

22. An adaptive histogram equalization processor according to claim 21, wherein the structure determining means is adapted to delete single and sparsely distributed edge points determined by the edge detection method and/or is adapted to connect broken edges by means of e.g. morphological filters.

23. An adaptive histogram equalization processor according to claim 14, further comprising a skin color detection device for determining skin colored pixels, whereby for regions with skin colored pixels the histogram building device is adapted to modify the histogram based on the amount of skin colored pixels, whereby the larger the number of skin colored pixels in the region the higher the degree of balance or of uniform distribution of the histogram obtained by the histogram building device.

24. An adaptive histogram equalization processor according to claim 14, wherein the histograms are homogenized.

25. An adaptive histogram equalization processor according to claim 14, wherein the regions are overlapping and wherein the mapping device is adapted to map the values of pixels located in the non-overlapping part of a region and the values of pixels located in the overlapping part of a region only once.

26. An adaptive histogram equalization processor according to claim 14, wherein the pixel values are values relating to the brightness and/or colorness of the pixels, for example the luminance value and/or chrominance or chrominance difference signals of a YUV, YIQ, or YCbCr or other color model based image format.

27. A software product comprising program information that, when executed in a computing device, is operable to carry out the adaptive histogram equalization method according to any one of the claims 1 to 13.