Video sequence processing of pixel-to-pixel dissimilarity values

1. In a video processor, a method of analyzing an array of pixel-to-pixel dissimilarity values, representing dissimilarity between the pixels of a first image and the pixels of a second image, to identify a pixel which has a low pixel-to-pixel dissimilarity value and which has neighboring pixels which have a low pixel-to-pixel dissimilarity value; the method comprising the steps of:

obtaining a picture spatial activity measure which is a measure of a spatial activity in the first image;

filtering the pixel-to-pixel dissimilarity values with a filter having a filter aperture; and

varying the spatial extent of the effective filter aperture in dependence upon said picture spatial activity measure,

wherein the dissimilarity values are rectified displaced frame difference values, and

wherein the outputs of two or more filters of pixel dissimilarity values are combined in dependence on a picture activity measure.

2. In a video processor, a method of analyzing an array of pixel-to-pixel dissimilarity values, representing dissimilarity between the pixels of a first image and the pixels of a second image, to identify a pixel which has a low pixel-to-pixel dissimilarity value and which has neighboring pixels which have a low pixel-to-pixel dissimilarity value; the method comprising the steps of:

obtaining a picture spatial activity measure which is a measure of a spatial activity in the first image;

filtering the pixel-to-pixel dissimilarity values with a filter having a filter aperture; and

varying the spatial extent of the effective filter aperture in dependence upon said picture spatial activity measure,

wherein the dissimilarity values are rectified displaced frame difference values, and wherein picture activity values and dissimilarity values are separately accumulated within a common measurement window of increasing size, and the filter output is the accumulated dissimilarity value when the accumulated activity value reaches a threshold value.

3. A method according to claim 2 in which the contribution of pixels to the said accumulation of dissimilarity values depends on a candidate filter aperture.

4. A method according to claim 2 in which accumulated dissimilarity values and accumulated activity values are calculated for a common measurement window comprising a non-integral number of pixels by interpolation between respective values derived from respective common measurement widows comprising integral number of pixels.

5. A method according to claim 4 in which the filtered output is the interpolated dissimilarity value for which the interpolated activity value is equal to the said threshold.

6. A method according to claim 1 in which the filter output is reduced in proportion to the number of pixels contributing to the accumulated dissimilarity value.

7. Video sequence processing apparatus for analyzing motion-compensated pixel-to-pixel dissimilarity values, the apparatus comprising:

a spatial filter adapted to receive motion-compensated pixel-to-pixel dissimilarity values, the spatial filter having a variable effective filter aperture; and

a picture spatial activity measurer adapted to receive a current picture and to provide a measure of spatial activity in the current picture;

in which the spatial extent of the effective filter aperture is varied in dependence on said picture spatial activity measure;

in which the dissimilarity values are rectified displaced frame difference values;

in which said spatial filter comprises two or more filters and in which the spatial extent of the effective filter aperture is varied by combining the outputs of said two or more filters in dependence on the picture activity measure.

8. A non-transitory computer readable storage medium storing a computer program adapted to cause programmable apparatus to implement a method of analyzing an array of pixel-to-pixel dissimilarity values to identify a pixel which has a low pixel-to-pixel dissimilarity value and which has neighboring pixels which have a low pixel-to-pixel dissimilarity value; the method comprising the steps of obtaining a picture spatial activity measure by measuring spatial activity in a single image; filtering the pixel-to-pixel dissimilarity values with a filter having a filter aperture; and varying the spatial extent of the effective filter aperture in dependence upon said picture spatial activity measure, in which the dissimilarity values are rectified displaced frame difference values, in which the outputs of two or more filters of pixel dissimilarity values are combined in dependence on a picture activity measure.

9. A medium according to claim 8 in which picture activity values and dissimilarity values are separately accumulated within a common measurement window of increasing size, and the filter output is the accumulated dissimilarity value when the accumulated activity value reaches a threshold value.

10. A medium according to claim 9 in which the contribution of pixels to the said accumulation of dissimilarity values depends on a candidate filter aperture.

11. A medium according to claim 9 in which accumulated dissimilarity values and accumulated activity values are calculated for a common measurement window comprising a non-integral number of pixels by interpolation between respective values derived from respective common measurement widows comprising integral number of pixels.

12. A medium according to claim 11 in which the filtered output is the interpolated dissimilarity value for which the interpolated activity value is equal to the said threshold.

13. A medium according to claim 8 in which the filter output is reduced in proportion to the number of pixels contributing to the accumulated dissimilarity value.

14. A method according to claim 1 in which the spatial extent of the effective filter aperture is varied such that in areas of lower spatial activity, a larger filter aperture is employed and that in areas of higher spatial activity, a smaller filter aperture is employed.

15. A method according to claim 2 in which the spatial extent of the effective filter aperture is varied such that in areas of lower spatial activity, a larger filter aperture is employed and that in areas of higher spatial activity, a smaller filter aperture is employed.