Abstract:Usually sparse representation of signal is not unique, which results in a large number of sparse representation algorithms. An improved Orthogonal Matching Pursuit(OMP) algorithm is proposed. The atoms are selected more quickly with nonlinear decline threshold and the set of alternative atoms is determined, which improves the algorithm speed. Regularized secondary screening can remove lower-energy atoms from the alternative atoms set to ensure the accuracy. A stop condition for iteration is preset to realize the adaptive sparsity of new algorithm. Simulation results show that, the improved algorithm can keep a balance between accuracy and speed for sparse solving with a faster speed than Basis Pursuit(BP) algorithm and a higher accuracy than OMP, Regularized OMP(ROMP) and Backtracking-based Adaptive OMP(BAOMP) algorithms.