Piecewise smooth signal x, D is the differential matrix.
Suppose x is corrupted by noise e, b is the observation b=x+e.
Image denoise
L1-minimization(Total Variation) is used to recovery x and preserve sharp features in x.
results:
Where H is the convolution matrix.
results:
Some
variation:
Multi-dimensional extension:
The function value is a vector of size dimx1. TV-norm can either be the anisotropic TV norm or the isotropic norm.
Results of color image smooth, curve, point cloud denoise.
Image enhancement
: