Noise reduction systems and methods
View Patent ↗Embodiments of noise reduction systems and methods are disclosed. One method embodiment, among others, comprises transforming a residual signal to produce transform coefficients, and applying the transform coefficients to a quantization matrix and a configurable gain matrix to provide a frequency-selective weighting of the transform coefficients having a gain that is sensitive to noise levels.
1. A noise reduction method, comprising:
processing input video data to produce a residual signal;
transforming the residual signal to produce transform coefficients; and
applying the transform coefficients to a quantization matrix and a configurable gain matrix to provide a frequency-selective weighting of the transform coefficients having a gain that is responsive to noise levels of the input video data; and
outputting the frequency-selective weighting of the transform coefficients as output video data.
2. The method of claim 1 , further including the steps of
determining a motion level through a comparison of pixels in two temporally different image regions; and
setting a gain of the configurable gain matrix as a function of the determined motion level, wherein the two temporally different image regions are connected by a motion vector and transforming comprises 2-dimensional discrete cosine transforming.
3. The method of claim 1 , wherein applying comprises multiplying the transform coefficients element by element by the quantization matrix to produce products, wherein the products are divided element by element by the configurable gain matrix to produce quotients.
4. The method of claim 1 , further comprising receiving configurable gain and shape parameters to weight the configurable gain matrix.
5. The method of claim 1 , further comprising manually adjusting at least one of the gain parameters and the shape parameters.
6. The method of claim 1 , further comprising automatically adjusting the gain parameters based on the noise information.
7. A noise reduction system, comprising:
an encoder comprising quantization logic configured with a quantization matrix and a gain matrix that is configurable based on gain and shape parameters, the quantization logic configured to apply transform coefficients produced from a residual of input video data to the quantization matrix and the gain matrix to generate a frequency-selective weighting of the transform coefficients having a gain that is responsive to noise levels of the input video data.
8. The system of claim 7 , wherein the encoder module further comprising a transform module configured to transform the residual to the frequency domain.
9. The system of claim 7 , further comprising a motion estimation module configured to generate noise information based on noise level estimates and motion detection.
10. The system of claim 9 , wherein the quantization logic is configured to receive the noise information and adjust the gain parameter based on the noise information.
11. The system of claim 7 , wherein the quantization logic is configured to receive the gain and shape parameters as adjusted by a user.
12. The system of claim 7 , wherein the encoder module is configured in hardware.
13. The system of claim 7 , wherein the encoder module is configured in software.
14. The system of claim 7 , wherein the encoder module is configured in software and hardware.
15. A noise reduction system, comprising:
means for processing input video data to produce a residual signal;
means for transforming the residual signal to produce transform coefficients; and
means for applying the transform coefficients to a quantization matrix and a gain matrix adjustable based on configurable gain parameters and shape parameters to provide a frequency-selective weighting of the transform coefficients having a gain that is dependent on noise levels of the input video data.
16. The system of claim 15 , wherein the means for transforming comprises a discrete cosine transform module.
17. The system of claim 15 , wherein the means for applying comprises a quantization module having a quantization matrix and the gain matrix.
18. The system of claim 15 , further comprising means for providing noise level and motion level estimates.
19. The system of claim 15 , further including
means for determining a motion level through a comparison of pixels in two temporally different image regions; and
means for setting a gain of the configurable gain matrix as a function of the determined motion level, wherein the two temporally different image regions are connected by a motion vector.
20. The system of claim 19 , further including means for providing noise level and motion level estimates, and wherein the means for transforming includes a discrete cosine transform module and the means for applying includes a quantization module having a quantization matrix and the gain matrix.
21. The method of claim 2 , further including the steps of:
receiving configurable gain and shape parameters to weight the configurable gain matrix;
manually adjusting at least one of the gain parameters and the shape parameters; and
automatically adjusting the gain parameters based on the noise information.
22. The system of claim 7 , further including
logic for determining a motion level through a comparison of pixels in two temporally different image regions; and
logic for setting a gain of the configurable gain matrix as a function of the determined motion level, wherein the two temporally different image regions are connected by a motion vector and the quantization logic is further configured to transform the coefficients using 2-dimensional discrete cosine transforming.
23. The system of claim 22 , further including
a transform module configured to transform the residual to the frequency domain; and
a motion estimation module configured to generate noise information based on noise level estimates and motion detection; and wherein the quantization logic is configured to receive the noise information and to adjust the gain parameter based on the noise information and to receive the gain and shape parameters as adjusted by a user.