Block noise detection and filtering

1. A method of detecting block noise in an image, the method, comprising:

computing difference magnitudes in pixel values for adjacent pixels in the image;

eliminating a set of difference magnitudes from the difference magnitudes resulting from features in the image;

applying a median filter to the image to generate an output;

subtracting the output from the difference magnitudes;

performing edge detection in the image to detect edge pixel locations.

2. The method of claim 1 , wherein, the difference magnitudes comprise horizontal difference magnitudes for horizontally adjacent pixels and vertical difference magnitudes for vertically adjacent pixels;

further comprising,

averaging the horizontal difference magnitudes in a vertical direction to generate a first vector of average values of the horizontal difference magnitudes of a width of the image;

averaging the vertical difference magnitudes in a horizontal direction to generate a second vector of average values of the vertical difference magnitudes of a height of the image.

3. The method of claim 2 , further comprising,

plotting the first vector against pixel locations along the width of the image in a graph.

4. The method of claim 2 , further comprising, plotting the second vector against pixel locations along the height of the image in a graph.

5. The method of claim 2 , further comprising, removing a DC level from the average values of the vertical difference magnitudes or the average values of the horizontal difference magnitudes.

6. The method of claim 5 , wherein, the DC level is removed by applying a median filter to the average values and subtracting the result from the average values.

7. The apparatus of claim 5 , wherein, the optical disk reader is an HD-DVD reader.

8. The method of claim 2 , further comprising, suppressing values in the first and

second vectors near the edges of the image.

9. The method of claim 2 , further comprising,

determining normalized sums of the difference magnitudes for a set of candidate interval sizes and a set of candidate offsets;

using a largest value of the normalized sums to determine an interval size and offset of the block noise.

10. The method of claim 9 , further comprising, using the interval size and offset to determine a nominal block size and a starting boundary block location.

11. The method of claim 10 ,

further comprising, identifying a subsequent boundary block location;

wherein, the subsequent boundary block location is determined by:

incrementing the starting block location by the nominal block size;

identifying a location of a largest or near-largest difference magnitude value in a window around the subsequent boundary block location, and

selecting the location of the largest or near-largest value as the subsequent block boundary location.

12. The method of claim 11 , further comprising, generating a block boundary map using multiple subsequent boundary block locations.

13. The method of claim 9 , further comprising,

determining three normalized sums of the difference magnitudes;

wherein, in a first summation of the three normalized sum, computing sum values using a nominal location and the nominal location −1 and selecting the location that yields a larger of the sum values;

wherein, in a second summation of the three normalized sum, computing sum values using a nominal location and the nominal location +1 and selecting the location that yields a larger of the sum values;

wherein, in a third summation of the three normalized sum, computing a sum value using the nominal location selecting the nominal location.

14. The method of claim 13 , further comprising,

identifying a candidate interval size and offset value which generates a max normalized sum of difference magnitude;

compare the max normalized sum to sums computed for the other two offset values of the candidate interval size to determine a block size of the block noise and to select a final block size and offset value using the nominal block size and the starting boundary location.

15. The method of claim 9 ,

further comprising, using normalized sums to determine a set of noise characteristics of the block noise and using difference magnitude data to determine a set of image characteristics of the image.

16. The method of claim 15 ,

wherein, the set of noise characteristics include, strength of block noise and block size;

wherein, the set of image characteristics include, amount of detail in the image.

17. The method of claim 15 , further comprising, determining a gain value for a block noise filter using the set of noise and image characteristics.

18. The method of claim 2 , wherein, the suppressing the edge difference magnitudes at the edge pixel locations further comprises:

setting the edge difference magnitude to zero at a location of the edge pixel locations where a corresponding gradient magnitude equals or exceeds a gradient threshold value.

19. The method of claim 2 , wherein, the difference magnitudes are computed for luma and chroma.

20. The method of claim 19 , further comprising, generating a chroma block boundary map using the difference magnitudes computed for the chroma and a luma block boundary map using the difference magnitudes computed for the luma.

21. The method of claim 1 , further comprising, using the difference magnitudes to determine whether the image is scaled.

22. The method of claim 1 , wherein, the horizontally and vertically adjacent pixels are of I-pixel differentials.

23. The method of claim 1 , further comprising, attenuating the set of difference

magnitudes by identifying the set of difference magnitudes equal to or exceeding a threshold value and setting the set of difference magnitudes to the threshold value.

24. The method of claim 1 , wherein, the subtracting the output of the median filter from the difference magnitudes further comprises: clipping values less than zero at zero.

25. The method of claim 1 , wherein, the edge detection is performed using Sobel filters.

26. The method of claim 25 , wherein, the Sobel filters are 3×3.