Filter a 2D tensor field with the normalized convolution version of the domain transform filter.
Filter a 2D tensor field with the normalized convolution version of the domain transform filter. This smoothes textures while preserving edges. It is similar to a bilateral filter, but much faster.
The color field to be filtered.
Spatial extent of nonlinear filter.
Range extent of nonlinear filter (same concept as range in bilateral filters).
Number of iterations to run filter. Usually converges in 10 to 15 iterations, but 3 iterations is good enough for most applications.
Filtered version of input.
Implements the normalized convolution version of the "domain transform" filter. This filter is "edge-aware" and able to filter without blurring edges. For a complete description, see "Domain transform for edge-aware image and video processing," Gastal and Oliveira, 2011.