Font Size: a A A

Study On Image Reconstruction Algorithm Of Compressed Sensing Based On Approximation Zero Norm

Posted on:2018-01-17Degree:MasterType:Thesis
Country:ChinaCandidate:Y L GongFull Text:PDF
GTID:2348330533463225Subject:Computational Mathematics
Abstract/Summary:PDF Full Text Request
With rapid development of communication technology,information content is also increasing greatly.All kinds of complex signals have emerged in life of human society.Human life more and more depends on signal process technology.As one of the most basic theories to process signals,compressed sensing theory has penetrated into many fields,such as informatics,biology science,medicine and so on.In recent years,signal reconstruction has attracted wide interests of researchers.The study of signal reconstruction based on compressed sensing is one of the representative issues.Compressed sensing makes it possible to sample and compress signals at the same time and it makes signal processing more concise and effective.Therefore,it is of great practical significance and theoretical value to study the problem of signal reconstruction based on compressed sensing.In this paper,we do some research on signal reconstruction algorithm of compressed sensing based on approximation zero norm.The key to algorithm is to construct the appropriate function to approximate 0l norm,and then the problem is transformed into an optimization problem.It can be easily solved through effective algorithm.Taking this as a starting point,the main research work is summarized as follows:Firstly,a new simple fractional function is proposed to approximate 0l norm.It not only ensures the accuracy,but also reduces the computational complexity of the algorithm with simple expression in the iterative calculation.The simulation results also verify the effectiveness of the proposed function.Secondly,the problem of one-dimensional sparse signal reconstruction algorithm based on approximate 0l norm is studied.The solution is obtained by solving an unconstrained problem.In order to ensure the direction of Newton is reduced,the Newton algorithm is applied directly to the unconstrained problem.As a result,the error is minimized as much as possible.Compared with the existing similar algorithms,the proposed algorithm has an advantage in accuracy.Finally,the realization of image reconstruction algorithm is studied.The redundant ridge transform matrix is used to represent the image as the basic functions.Reference to the idea of block compressed sensing,it is only need to reconstruct each image block for the whole image.The experiment results show the effectiveness of the algorithm proposed.
Keywords/Search Tags:Compressed sensing, image reconstruction, Newton algorithm, 0l norm
PDF Full Text Request
Related items