Skip to main content

Table 1 Simultaneous Orthogonal Matching Pursuit (SOMP) algorithm

From: An improved sparse representation model with structural information for Multicolour Fluorescence In-Situ Hybridization (M-FISH) image classification

Algorithm 1: SOMP
(1): Input: training sample matrix A , testing sample matrix Y
(2): Output: Row-wise sparse solution X ^
(3): Initialization: residual R 0 = Y , X ^ 0 =0, non-zero rows Ω=, i = 0
(4): While stopping criterion false do
1). Find a new atom from matrix A to best approximate the current residual based on q-norm: w= arg  max k Ω a k T R i - 1 q
2). Update the non-zero row support Ω=Ω w .
3). Update the signal estimation X ^ i = A Ω T A Ω + A Ω T Y , where A Ω denotes the sub-matrix of A consisting of the atoms from matrix A, and the residual: R i = Y - A Ω X ^ i .
4).i = i + 1.
(5): End while
(6): Return: X ^ = X ^ i