Abstract:

According to advantages and drawbacks of existing measurement matrixes, a local randomized Hadamard matrix is adopted, which has high correlativity, stochastic independence and fast computation capability. Meanwhile, according to the question which is hard to reconstruct stably original signals for the standard orthogonal matching pursuit (OMP) algorithm when the samples are compressible signals or are contaminated with noises, a novel orthogonal multimatching pursuit algorithm based on local randomized Hadamard matrix is proposed. The proposed algorithm can precisely reconstruct signals by using the special structural features of the local randomized Hadamard matrix. The simulation results of one dimensional signals and real image show that the proposed algorithm is superior to other greedy algorithms and convex basis pursuit (BP) method no matter whether the samples are contaminated with noise or not.