Skip to main content

Table 1 Summary of the K-cluster-valued CoSaMP algorithm

From: K-cluster-valued compressive sensing for imaging

- Input: Measurement matrix Φ, measurement vector y, sparsity level S, and intensity cluster number K.
- Output: S-sparse approximation x ^ of target image x.
- Initialization: x ^ 0 = 0 , r= y and i = 1.
  While halting criterion = true
     1 zΦ*r{Compute the proxy of residual}
     2 Ω ← supp(z2K) {Identify the largest 2K components of the proxy}
     3 TΩ s u p p ( x ^ ( i - 1 ) ) {Merge supports}
     4 b | T Φ T y and b | T C 0 {Estimate the image by least-squares solution}
     5 x ^ i b K {Prune to obtain the image approximation for the next iteration or output}
       While n = 1 N k = 1 K r n k x ^ n i - μ k i 2 <threshold
          6 For each n = 1, 2,..., N, {Assign intensity values of each pixel to the closest intensity cluster}
                 r n k i = 1 i f k = arg min j x ^ n i - μ k i 2 a n d x ^ n i 0 0 o t h e r w i s e
          7 For each n=1,2,,N, μ k i = Σ n = 1 N r n k i x ^ n i Σ n = 1 N r n k i . {Obtain the K cluster centres}
       End
   8 For n = 1, 2,..., N, if r nk = 1, then x ^ n i = μ k i {Optimize the estimated image with K-cluster intensities}
   9 r=y-Φ x ^ i {Update the measurement residual for the next iteration}
   10 i = i + 1 {Update the iteration number}
  End
  return x ^ x ^ i