Image Reconstruction Of Compressive Sensing With Alternative Learning And Immune Optimization

With the advent of the information age, the demand for information is getting increasing. However, Traditional Nyquist Sampling theorem requires sampling rate of signal no less than twice the signal bandwidth, which is impossible for the ability of signal processing. Compressed Sensing(CS) theory is a great change in the field of information processing in recent years. It points out that if the signal is sparse or compressible in an orthogonal basis or tight framework, we can sample at the rate of far below the Nyquist, and reconstruct the original signal accurately. Currently, Compressed Sensing theory has a broad prospect in compact imaging system, analog/information conversion, biologic sensing and other fields. This paper mainly studies image reconstruction problem of compressed sensing, and a new image reconstruction framework based on alternative learning and immune optimization is proposed under the traditional immune optimization framework, The innovative aspects of this paper is as follows:1.This paper propose ideas of the alternative learning by using the locations of the sparse coefficients and the values of the sparse coefficient. Not only how to determine the locations of the sparse coefficient are concerned, the values of the sparse coefficient are solved but also in the guidance of the locations, and then a compressed sensing reconstruction framework via filtering operator and alternative optimization is established. This framework can accurately locate the position of large coefficients of wavelet domain, and also compute the coefficient values, thus reconstruct an image with small error and good visual effects.2.In this paper, We put the adaptive filtering as an immune operator into our immune optimization framework. This operator can accurately capture the details of an image, especially the texture and edge regions; therefore, the quality of reconstruction image is improved.3.In this paper, projection into convex set is used in wavelet domain, and is put into the immune optimization as an immune operator, which makes the results of each iteration near the solution of the problem, and also improve the performance of the algorithm.Finally, we simulate the proposed reconstruction algorithm in Matlab software. The simulation results show the validity of the algorithm, which provides new ideas and inspiration for subsequent compressed sensing reconstruction algorithm. Experimental results show that the proposed algorithm is better than comparison algorithm both in image visual effect and in numerical error.