Method and apparatus for MPEG artifacts reduction

1. A method of reducing mosquito noise in an image frame of a video signal, the method comprising:

examining a window of the image frame, the window including a plurality of pixel values;

subtracting a selected one of the pixel values from the pixel values in the window to define a modified window of pixel value differences;

computing a local mean value E DC of the pixel value differences in the modified window of pixel value differences;

applying a highpass filter to the pixel value differences in the modified window of pixel value differences to generate a reducing value E cb ;

computing a weighting coefficient α as a function of a variance of the pixel value differences in the modified window of pixel value differences;

computing a weighted sum of the reducing value E cb and the local mean value E DC after multiplication with the weighting factor α to define a mosquito noise reduction value u according to the relation:

u=αE cb +(1−α) E DC ;

generating a nonlinear function Ψ(u) of the difference between the mosquito noise reduction value u and the local mean value E DC ;

computing an artifact attenuation factor λ as a function of a dynamic range measure of the pixel value differences in the modified window of pixel value differences;

adding the local mean value E DC to Ψ(u);

multiplying the artifact attenuation factor λ by the sum of the local mean value E DC and Ψ(u); and

adding the selected one of the pixel values to the product of the artifact attenuation factor λ and the sum of the local mean value E DC and Ψ(u).

2. The method of claim 1 further comprising:

limiting the pixel value differences in the modified window of pixel value differences by a threshold.

3. The method of claim 1 , wherein computing the local mean value E DC includes:

applying a 2D lowpass filter to the pixel value differences in the modified window of pixel value differences.

4. The method of claim 1 , wherein:

the selected pixel value is a central pixel of the window.

5. The method of claim 1 , wherein:

the window is a n×n window of pixels.

6. The method of claim 5 , wherein:

n is a small odd number.

7. The method of claim 6 , wherein:

n=3.

8. An apparatus for reducing mosquito noise in an image frame of a video signal, the apparatus comprising:

a windowing module configured to examine a window of the image frame, the window including a plurality of pixel values;

a subtraction module configured to subtract a selected one of the pixel values from the pixel values in the window to define a modified window of pixel value differences;

a local mean determination module configured to compute a local mean value E DC of the pixel value differences in the modified window of pixel value differences;

a mosquito noise reduction module configured to generate a mosquito noise reduction value u for the pixel value differences in the modified window of pixel value differences,

wherein the mosquito noise reduction module includes:

a highpass filter configured to generate a reducing value E cb when applied to the pixel value differences in the modified window of pixel value differences;

a variance module configured to compute a weighting coefficient α as a function of a variance of the pixel value differences in the modified window of pixel value differences; and

a summing module configured to compute a weighted sum of the reducing value E cb and the local mean value E DC to define the mosquito noise reduction value u according to the relation:

u=αE cb +(1−α) E DC ;

a nonlinear processing module configured to generate a nonlinear function ψ(u) of the difference between the mosquito noise reduction value u and the local mean value E DC ;

a smooth attenuation module configured to compute an artifact attenuation factor λ as a function of a dynamic range measure of the pixel value differences in the modified window of pixel value differences; and

an output module configured to:

add the local mean value E DC to ψ(u);

multiply the artifact attenuation factor λ by the sum of the local mean value E DC and ψ(u); and

add the selected one of the pixel values to the product of the artifact attenuation factor λ and the sum of the local mean value E DC and ψ(u).

9. The apparatus of claim 8 , wherein the mosquito noise reduction module includes:

a limiting module configured to limit the pixel value differences in the modified window of pixel value differences by a threshold.

10. A computer readable medium encoded with a computer software and when executed operable to:

subdivide an image frame of a video signal to generate a window of the image frame, the window including a plurality of pixel values;

subtract a selected one of the pixel values from the pixel values in the window to define a modified window of pixel value differences;

compute a local mean value E DC of the pixel value differences in the modified window of pixel value differences;

compute a mosquito noise reduction value u based on the pixel value differences, wherein computing the mosquito noise reduction value includes:

applying a highpass filter to the modified window of pixel value differences to generate a reducing value E cb ;

computing a weighting coefficient α as a function of a variance of the pixel value differences in the modified window of pixel value differences; and

computing a weighted sum of the reducing value E cb and the local mean value E DC , after multiplication with the weighting factor α, to define the mosquito noise reduction value u according to the relation:

u=αE cb +(1−α) E DC ;

generate a nonlinear function ψ(u) of the difference between the mosquito noise reduction value u and the local mean value E DC ;

compute an artifact attenuation factor as a function of a dynamic range measure of the pixel value differences in the modified window of pixel value differences;

add the local mean value E DC to ψ(u);

multiply the artifact attenuation factor by the sum of the local mean value E DC and ψ(u); and

add the selected one of the pixel values to the product of the artifact attenuation factor and the sum of the local mean value E DC and ψ(u).

11. The computer program product software of claim 10 , further operable to:

limit the pixel value differences in the modified window of pixel value differences by a threshold.

12. The software of claim 10 , wherein computing the local mean (DC) value includes:

applying a 2D lowpass filter to the pixel value differences in the modified window of pixel value differences.

13. The software of claim 10 , wherein:

the selected pixel value is a central pixel of the window.

14. The software of claim 10 , wherein:

the window is a n×n window of pixels.

15. The software of claim 14 , wherein:

n is a small odd number.

16. The software of claim 15 , wherein:

n=3.