Reseach On Recovery Algorithms Of Compressive Sensing

Compressive Sensing is a new sampling theory, it takes advantage of the sparsesignal, using the random sampling matrix to get the sample in the case of less than theNyquist sampling rate. Then it reconstructs the signal perfectly with non-linearreconstruction algorithm. Compressive Sensing is mainly applicated in the field ofinformation theory, image processing, pattern recognition and wireless communication.The calculation of signal reconstruction algorithm is usually large, so it is difficultto meet real-time requirements. Intelligent algorithm has a huge advantage to solvenon-linear optimization problems. In the thesis, we research on a part of intelligentalgorithms and fuse the OMP reconstruction algorithm with the intelligent algorithms,propose the OMP fusion algorithm with Artificial bee colony algorithm and adaptivequantum genetic algorithm respectively.Specifically, the mainly research works and contributions as follows:1. Based on the signal sparse decomposition problem, The paper improved thetraditional iteration termination conditions to choose the appropriate terminate thresholdand proposed the fusion ABC-OMP algorithm. Experimental results show that thefusion algorithm is equal to particle swarm algorithm and genetic algorithm in quality,but is better than particle swarm algorithm and genetic algorithm in speed.2. An adaptive quantum genetic algorithm on Bloch sphere is proposed based onthe quantum genetic algorithm on Bloch coordinates of qubits. The algorithm uses twoways to select a part of the Bloch sphere for searching. The paper proved that the twomethods are able to contain all the solutions of the continuous optimization problem intheory, and proposed a method of approximately equal-area to search the selected Blochsphere, and derived the inverse relationship between the two-phase. The simulationresults show that the approach is equal to Quantum genetic algorithm on Blochcoordinates of qubits in search capability, but the optimization efficiency is significantlyimproved. Finally the OMP algorithm is fused with ABQGA algorithm.3. In this paper, The simulation results show that the proposed fusion algorithms iseffective both for one-dimensional digital signal and two-dimensional image signal.