infimalConvolutionFilter
matrix-valued image with the TV-norm. \argmin_u{\|u-u_0\|_2^2 + \lambda \|\nabla u\|_{2,1} + \mu \|H u\|_{2,1}}, \lambda,\mu > 0 Written by Johannes Persch, Ronny Bergmann, 09.06.2015 function ma_val_im_reg_2nd_ord minimzes the second order model of a OUTPUT: u_hat: computed minimizer diff: differences of u and v between two successive iterations
12.05.15 by Markus Loeckel 25.05.15 substituted gauss seidel by mldivide 26.05.15 modified SPD by not deleting half of the entries 30.06.15 projection onto sphere added 01.07.15 mask added 14.07.15 alternative minimzation w.r.t. u via preconditioned cg added 16.07.15 first-and second-order combined 16.07.15 stopping criterion added
Syntax
MTEX 5.0.0 |