Hand Depth Image Denoising and Superresolution via Noise-Aware Dictionaries

Algorithm 3

The noise-aware orthogonal matching pursuit.

  input: vector , dictionary , sparsity , residual threshold , noise index threshold

  output: sparse coefficient of with respect to

() Initialize residual , index set , ;

() for    do

()   Denote , to be all columns of ;

()   ;

()   Update the index set ;

()   Update the atom matrix ;

()   ;

()   Update residual ;

()   ; / collect all indices of great components of residual   /

()   if    then / when has only a small number of great components, we re-update the atom by removing those

     components from /

()     foreach    do  ;

()     ;

()     Re-update the index set ;

()     Re-update the atom matrix ;

()     ;

()     Re-update residual ;

()   end

() end