Research On Compressed Sensing Reconstruction Algorithm Based On Intelligent Optimization Algorithm

With the continuous advancement of social information,electronic technology has been developing rapidly.Due to the performance of the sensor is more and more perfect,large amounts of data is provided for the system.More and more data is constantly pounding each domain.There is a public problem how to more effectively express huge amounts of data.Compressed sensing theory broke the Nyquist sampling theorem.The theory merged the collection and compression process.In the process of data acquisition and processing,it can greatly reduce the burden of storage and transmission.It provided a novel approach.Intelligent optimization algorithm can take advantage of empirical information and the ability of fine local search to solve the optimization problem.Particle swarm optimization algorithm is a kind of intelligent optimization algorithm which is frequently used.Particle swarm optimization algorithm has many advantages,such as simple operation,easy to code and less parameters,etc.Due to compressed sensing theory reconstruction algorithm is to solve the optimization problem.Therefore,this thesis mainly optimized and improved several typical matching pursuit algorithms with the particle swarm optimization algorithm.In this thesis,the main research content are as follows:(1)The greedy algorithm is studied.The several typical matching pursuit algorithm are summarized and compared in detail.The characteristics and process of these algorithms is summarized in detail.Through a large number of simulation experiment,these algorithms are compared and analyzed.(2)If the setting parameters of stage-wise orthogonal matching pursuit algorithm and sparsity adaptive matching pursuit algorithm is unreasonable,it don't make the reconstruction algorithm achieve the best effect.An improved design is proposed for the problem that the effect of two kinds of reconstruction algorithm is easily affected by the parameters.The design is base on particle swarm optimization algorithm.And the optimization process flow chart is given.At last,its feasibility and effectiveness are verified by a large number of simulation experiments.(3)A novel improved method is proposed for the compressive sampling matching pursuit algorithm.There are many improved methods mainly improved the process that select the best matched atom.The solution of algorithm is not optimal solution but suboptimum.Previous improved methods do not improve this defect.Aiming at the defect of the algorithm,an improved compressive sampling matching pursuit algorithm is proposed.The sparse coefficient solution is learned and optimized by using particle swarm optimization algorithm.The sparse coefficient solution is closer to the original solution.To a certain extent,the defect is made up.At last,the feasibility and effectiveness of improved scheme are validated by a large number of simulation experiments.