Automatic parameter control for spatial-temporal filter
A method for image processing includes receiving an input image and estimating a noise characteristic of the input image, wherein the noise characteristic includes a temporal characteristic and a spatial characteristic. Spatially filtering the input image based upon the spatial characteristic, wherein the spatial characteristic is dependent upon the temporal characteristic.
1. A method for image processing comprising:
(a) receiving an input image;
(b) estimating a noise characteristic of said input image, wherein said noise characteristic includes a temporal characteristic and a spatial characteristic;
(c) spatially filtering said input image based upon said spatial characteristic, wherein said spatial characteristic is dependent upon said temporal characteristic;
(d) wherein said temporal characteristic is based upon a motion compensation estimation between said input image and a temporally different image;
(e) wherein said temporally different image is a previous image of a sequence of images;
(f) wherein said temporal characteristic is further based upon an estimate of spatial noise in said input image;
(g) wherein said temporal characteristic is further based upon a previously filtered different image.
2. The method of claim 1 wherein said previously filtered different image is based upon a temporal characteristic measure of said previous different image.
3. The method of claim 2 wherein said previously filtered different image is maintained in a state buffer.
4. The method of claim 1 wherein a weighted temporal value is determined based upon said input image, temporally different image, and said temporal characteristic.
5. The method of claim 4 wherein said spatially filtering is different for different portions of said input image.