Method and device for adaptive noise measurement of a video signal

Patent metadata

Field	Value
Publication number	US-9444977-B2
Application number	US-201113190807-A
Country	US
Kind code	B2
Filing date	Jul 26, 2011
Priority date	Jul 26, 2010
Publication date	Sep 13, 2016
Grant date	Sep 13, 2016

Abstract

The present invention relates to method of adaptive noise measurement of a video signal comprising a sequence of images. The proposed method comprises the steps of searching one or more spatially uniform areas within a current image, determining a spatial uniformity value of said one or more spatially uniform areas and combining said spatial uniformity values to obtain a spatial noise value representing a first measure for the amount of noise in said current image, detecting motion by comparing the current image with one or more preceding images and/or one or more succeeding images of said sequence of images by use of a motion threshold to obtain one or more motion maps indicating the amount of motion between the current image and the respective preceding or succeeding image, searching one or more static areas within said one or more motion maps, determining a temporal uniformity value for said one or more static areas by comparing image values at positions in said one or more preceding images and/or said one or more succeeding images, at which a low amount or no motion is indicated in the respective one or more motion maps, said temporal uniformity value representing a measure for the amount of noise at said positions in said current image, and combining said obtained temporal uniformity values to obtain a combined noise value representing a second measure for the amount of noise in said current image.

First claim

The invention claimed is:

1. A method of adaptive noise measurement of an input video signal, said method comprising:
receiving the input video signal, the input video signal including a current image, one or more preceding images and one or more succeeding images;
searching for one or more spatially uniform areas within the current image;
determining a spatial uniformity value of said one or more spatially uniform areas and combining said spatial uniformity values to obtain a spatial noise value that represents a first measure for the amount of noise in said current image;
detecting motion by comparing the current image with the one or more preceding images and the one or more succeeding images by using the spatial noise value as a motion threshold;
obtaining one or more motion maps from said detected motion, said one or more motion maps indicating the amount of motion between the current image and the respective one or more preceding images or the one or more succeeding images;
searching for one or more static areas within said one or more motion maps;
determining, by circuitry, a temporal uniformity value for each of said one or more static areas in said one or more motion maps by comparing image values, at positions in said one or more preceding images and said one or more succeeding images, at which a low amount of motion or no motion is indicated in the respective one or more motion maps, and said temporal uniformity value representing a measure for the amount of noise at said positions in said current image; and
combining said determined temporal uniformity values to obtain a combined noise value representing a second measure for the amount of noise in said current image.

2. The method according to claim 1 , wherein said method is repeated for each image of said input video signal.

3. The method according to claim 1 , wherein said searching for one or more spatially uniform areas within said current image is performed by repeatedly selecting different areas of the current image and determining a spatial uniformity value for the selected area until one or more areas with a maximum uniformity or a uniformity below a spatial uniformity threshold are found.

4. The method according to claim 3 , wherein said searching for one or more spatially uniform areas within said current image uses one of a variance of an area, a standard deviation and the summed signal amplitudes, after high-pass filtering, of an area as a measure for spatial uniformity of said area.

5. The method of claim 3 , wherein said spatial uniformity value is normalized according to a size of the respective area.

6. The method according to claim 1 , further comprising:
verifying said obtained spatial noise value against a range of expected spatial noise values; and
discarding said obtained spatial noise value when said obtained spatial noise value is outside said range.

7. The method according to claim 1 , wherein said detecting motion uses said obtained spatial noise value or an adjustable threshold as motion threshold.

8. The method according to claim 1 , further comprising combining the obtained motion maps into a combined motion map by a pixel-wise operation that sets a pixel of said combined motion map to a value that indicates a low amount of motion or no motion when the same pixel in all motion maps indicates a low amount of motion or no motion, wherein
said combined motion map is used in said determining of said temporal uniformity values.

9. The method according to claim 1 , wherein said searching for one or more static areas within said one or more motion maps is performed by searching for one or more areas with a least amount of pixels having a value in said one or more motion maps indicating motion.

10. The method according to claim 1 , wherein said determining a temporal uniformity value uses one of a variance, a standard deviation and a summed difference of image values at said positions, at which a low amount of motion or no motion is indicated in the respective one or more motion maps.

11. The method according to claim 1 , wherein
said determining a temporal uniformity value is repeated for all positions, at which a low amount of motion or no motion is indicated in the respective one or more motion maps, and
said determined temporal uniformity values for a static area are combined, in particular summed up, to obtain a temporal uniformity value for said static area.

12. The method according to claim 1 , wherein
said determining a temporal uniformity value is repeated for further static areas, and
said determined temporal uniformity values for said static areas are combined to obtain said combined noise value by determining one of an average of all temporal uniformity values, a weighted average and a weighted median of all temporal uniformity values using as weight the number of non-motion or low-motion pixels in each static area.

13. The method according to claim 1 , further comprising:
verifying said determined temporal uniformity value against a range of expected temporal uniformity values; and
discarding said determined temporal uniformity value when said determined temporal uniformity value is outside said range.

14. The method according to claim 1 , further comprising:
comparing a number of positions of a static area, at which a low amount of motion or no motion is indicated in the respective one or more motion maps, against a range of expected numbers; and
discarding said respective static area when said number is outside said range.

15. The method according to claim 1 , further comprising:
comparing a number of positions of all static areas, at which a low amount of motion or no motion is indicated in the respective one or more motion maps, against a range of expected numbers; and
discarding said determined temporal uniformity values and using the obtained spatial noise value as measure for the amount of noise in said current image.

16. The method according to claim 1 , wherein the spatial uniformity value is weighted according to a pixel area of one or more the spatially uniform areas.

17. A device for adaptive noise measurement of an input video signal, said device comprising:
circuitry configured to
receive the input video signal, the input video signal including a current image, one or more preceding images and one or more succeeding images;
search for one or more spatially uniform areas within the current image;
determine a spatial uniformity value of said one or more spatially uniform areas and to combine said spatial uniformity values to obtain a spatial noise value that represents a first measure for the amount of noise in said current image;
detect motion by comparing the current image with the one or more preceding images and the one or more succeeding images by using the spatial noise value as a motion threshold;
obtain one or more motion maps from said detected motion, said one or more motion maps indicating the amount of motion between the current image and the respective one or more preceding images or the one or more succeeding images;
search for one or more static areas within said one or more motion maps;
determine a temporal uniformity value for each of said one or more static areas in said one or more motion maps by comparing image values, at positions in said one or more preceding images and said one or more succeeding images, at which a low amount of motion or no motion is indicated in the respective one or more motion maps, and said temporal uniformity value representing a measure for the amount of noise at said positions in said curr

Assignees

• Moesle Frank
• Springer Paul
• Schewior Gregor
• Sony Corp

Inventors

• Moesle Frank
• Springer Paul
• Schewior Gregor

Classifications

• H04N5/21 Primary

  Circuitry for suppressing or minimising disturbance, e.g. moiré or halo · CPC title

• H04N5/145

  Movement estimation (for video coding H04N19/51) · CPC title

• H04N17/004

  for digital television systems · CPC title