Method and system for removing artifacts in compressed images
View Patent ↗In a method for removing artifacts in a digital image that has been decoded from a compressed representation of the image, the pixels of the digital image are segmented into low detail pixels, high detail pixels, and boundary pixels, and then a single non-linear filtering method, such as a sigma filtering method, is applied to each pixel in the digital image. By selecting separate filter parameters for the low detail, high detail, and boundary pixels, the removal of artifacts is improved over prior methods.
1. A method for removing artifacts in a digital image that has been decoded from a compressed representation of the image, where the digital image is represented by a plurality of pixels, comprising the steps of:
segmenting the pixels of the digital image into low detail pixels, high detail pixels, and boundary pixels; and
applying a single filtering method to each pixel in the digital image, wherein separate filter parameters are selected for the filtering method for the low detail, high detail, and boundary pixels.
2. The method according to claim 1 , wherein the compressed representation of the image was produced using a discrete cosine transform.
3. The method according to claim 1 , wherein the compressed representation of the image was produced using a discrete wavelet transform.
4. The method according to claim 1 , wherein the single filtering method is a non-linear filtering method.
5. The method according to claim 4 , wherein the non-linear filtering method is implemented as a sigma filter.
6. The method according to claim 5 , wherein the filter parameters include sigma filter thresholds and sigma filter window sizes.
7. The method according to claim 1 , wherein the filter parameters include a dependence upon the compression level that was used in encoding the compressed representation of the image.
8. The method according to claim 7 , wherein the filter parameters for the low detail pixels depend only upon low frequency ones of said quantization parameters.
9. The method according to claim 7 , wherein the filter parameters for the boundary pixels depend only upon mid and high frequency ones of said quantization parameters.
10. The method according to claim 1 , wherein the filter parameters include a dependence upon a user-specified scaling factor.
11. The method according to claim 1 , wherein the filter parameters include a dependence on a resolution of the digital image.
12. The method according to claim 1 , wherein the filter parameters include a dependence on characteristics of a display device or media that is used to view the digital image.
13. The method according to claim 1 , wherein the step of segmenting includes comparing a local activity at each pixel to a predetermined threshold and labeling a pixel as a low detail pixel if the local activity is less than or equal to the threshold.
14. The method according to claim 1 , wherein the step of segmenting includes determining the boundary pixels as pixels that are located within a predetermined area of a predetermined number of low detail pixels.
15. A computer storage medium having instructions stored therein for causing a computer to perform the method of claim 1 .
16. A method for removing artifacts in a digital image that has been decoded from a compressed representation of the image, where the digital image is represented by a plurality of pixels each having a specific local activity, comprising the steps of:
segmenting the pixels of the digital image into a plurality of different categories based on the local activity of each of the pixels; and
applying a filter to each pixel in the digital image, wherein separate filter parameters are selected for the different categories of pixels, and the filter parameters include a dependence upon a characteristic of the compression that was used in encoding the compressed representation of the images.
17. The method according to claim 16 , wherein the compressed representation of the image was, produced using a discrete cosine transform.
18. The method according to claim 16 , wherein the compressed representation of the image was produced using a discrete wavelet transform.
19. The method according to claim 16 , wherein the step of segmenting the pixels includes segmenting the pixels into categories comprising at least low detail pixels and boundary pixels.
20. The method according to claim 19 , wherein the filter parameters, for a given level of compression, are additionally dependent upon a relationship between a value of the parameters for the low detail pixels and a value of the parameters for the boundary pixels, whereby the relationship constrains a difference between the values to a predetermined value.
21. The method according to claim 16 , wherein the filter parameters include a dependence upon compression level that was used in encoding the compressed representation of the image.
22. The method according to claim 21 , wherein the filter parameters for the low detail pixels depend only upon low frequency ones of said quantization parameters.
23. The method according to claim 21 , wherein the filter parameters for the boundary pixels depend only upon mid and high frequency ones of said quantization parameters.
24. The method according to claim 19 , wherein the step of segmenting the pixels includes additionally segmenting the pixels into high detail pixels.
25. The method according to claim 24 wherein the step of applying a filter comprises, besides filtering the low detail pixels and the boundary pixels, filtering only the high detail pixels that are located adjacent to a block boundary.
26. A computer storage medium having instructions stored therein for causing a computer to perform the method of claim 16 .
27. A system for removing artifacts in a digital image that has been decoded from a compressed representation of the image, where the digital image is represented by a plurality of pixels, said system comprising:
a first processing stage for segmenting the pixels of the digital image into low detail pixels, high detail pixels, and boundary pixels; and
a second processing stage for applying a single filtering method to each pixel in the digital image, wherein separate filter parameters are selected for the low detail, high detail, and boundary pixels.
28. The system according to claim 27 , wherein the compressed representation of the image was produced using a discrete cosine transform.
29. The system according to claim 27 , wherein the compressed representation of the image was produced using a discrete wavelet transform.
30. The system according to claim 27 , wherein the single filtering method is a non-linear filtering method.
31. The system according to claim 30 , wherein the non-linear filtering method is implemented as a sigma filter.
32. The system according to claim 31 , wherein the filter parameters include sigma filter thresholds and sigma filter window sizes.
33. A system for removing artifacts in a digital image that has been decoded from a compressed representation of the image, where the digital image is represented by a plurality of pixels each having a specific local activity, said system comprising:
a first processing stage for segmenting the pixels of the digital image into a plurality of different categories based on the local activity of each of the pixels; and
a second processing stage for applying a filter to each pixel in the digital image, wherein separate filter parameters are selected for the different categories of pixels, and the filter parameters include a dependence upon a characteristic of the compression that was used in encoding the compressed representation of the images.
34. The system according to claim 33 , wherein the compressed representation of the image was produced using a discrete cosine transform.
35. The system according to claim 33 , wherein the compressed representation of the image was produced using a discrete wavelet transform.
36. The system according to claim 33 , wherein the first processing stage segments the pixels into categories comprising at least low detail pixels and boundary pixels.
37. The system according to claim 36 , wherein the filter parameters, for a given level of compression, are additionally dependent upon a relationship between a value of the parameters for the low detail pixels and a value of the parameters for the boundary pixels, whereby the relationship constrains a difference between the values to a predetermined value.
38. The system according to claim 33 , wherein the filter parameters include a dependence upon a quantization level that was used in encoding the compressed representation of the image.
39. The system according to claim 38 , wherein the filter parameters for the low detail pixels depend upon low frequency quantization parameters.
40. The system according to claim 38 , wherein the filter parameters for the boundary pixels depend upon mid and high frequency quantization parameters.
41. A method for removing artifacts in a digital image decoded from a compressed image, the digital image having a plurality of pixels, the method comprising the steps of:
segmenting the pixels of the digital image into three groups: low
detail pixels, boundary pixels, and high detail pixels, each of said groups being exclusive of pixels of the other said groups; and
filtering said digital image, said filtering using different filter parameters for each of said groups.
42. The method of claim 41 wherein said segmenting further comprises:
segmenting the pixels of the digital image into low detail pixels and other pixels; and
segmenting said other pixels into boundary pixels and high detail pixels.
43. The method of claim 41 wherein said segmenting further comprises: determining a map of local activity values of said digital image, said local activity values each being associated with a respective said pixel;
assigning said pixels to said groups using said map.
44. The method of claim 43 wherein said segmenting further comprises assigning to said group of low detail pixels, ones of said pixels having respective said local activity values less than or equal to a predetermined threshold.
45. The method of claim 43 wherein said segmenting further comprises: determining a number of said low detail pixels within a
predetermined area of each of said other pixels to provide a low detail pixel number; and
assigning to said group of boundary pixels, ones of said other pixels having a respective said low detail pixel number greater than a predetermined number.
46. The method of claim 41 wherein said filtering further comprises filtering with a sigma filter and said filter parameters are sigma filter thresholds and sigma filter window sizes.
47. The method of claim 46 wherein the filter parameters for the low detail pixels depend only upon low frequency quantization parameters of the compressed image and the filter parameters for the boundary pixels depend only upon mid and high frequency quantization parameters of the compressed image.
48. A system for removing artifacts in a digital image decoded from a compressed image, the digital image having a plurality of pixels, the system comprising:
a first processing stage segmenting the pixels of the digital image into three groups: low detail pixels, boundary pixels, and high detail pixels, each of said groups being exclusive of pixels of the other said groups; and
a second processing stage filtering said digital image, said filtering using different filter parameters for each of said groups.