Output: A K-sparse approximation of the input signal;
Initialization: support sets , residual , iterations ,;
(1) Calculate based on DICE criteria, find the values with the largest measurement matrix and residuals by , which corresponding column vector with values greater than the threshold is recorded as set ;
(2) Get the candidate ,;
(3) Calculate , take the mark corresponding to the largest value into a collection , and calculate residual ;
(4) Determine , if meeting, executive 5, if not meeting, executive 6;
(5) Determine , if meeting, change the support set ,, executive 1
(6) Set ,, ,,;
(7) Determine , if meeting, executive 1, if not meeting, stop iterating and output , after cycle steps 1–6, it will output sparse approximation, namely