Research On A Radar Image Reconstruction Algorithm Based On Compressed Sensing

The traditional Nyquist sampling theorem requires that the sampling frequency of the signal should not be less than twice of the highest frequency in the analog signal spectrum,which brings a lot of difficulties to the cost of sampling and subsequent processing and transmission.In recent years,compressed sensing theory is born.For sparse signal or compressible signal,its sampling rate does not depend on the bandwidth of signal,but on the structure and content of information in the signal,so it can break through the bottleneck of Nyquist sampling theorem.At the same time,the theory can directly obtain compressed data while sampling,which can reduce the size and save storage spaceReconstruction algorithm is the core part in the theory of compressed sensing,which is also a research hot-spot at present.Therefore,based on the research of the overall framework of compressed sensing theory,this study mainly focus on the greedy iterative algorithm in re-construction algorithm.First the classical greedy algorithm,the orthogonal matching pursuit(OMP)and its improved algorithm the tree-based orthogonal matching pursuit(TOMP)al-gorithm are reviewed.Furthermore,on the basis of experiments,the core idea,framework,implementation process,the thesis discusses the advantages and disadvantages of the two algorithms and the reconstruction accuracy and other performance of the two algorithms.The thesis extends the application of TOMP algorithm to planetary sounding radar image.Firstly,the thesis extends the scope of application of the TOMP algorithm.Secondly,since the size of measurement matrix and sparse transformation matrix required for sampling and reconstruction of compressed sensing method is the square of image size,the direct appli-cation in planetary sounding radar image will increase the reconstruction time.The method of block sampling and reconstruction is applied to improve the calculation speed and the PSNR of the reconstructed image.The experimental results of different planetary sounding radar images clearly show that the PSNR and visual effect of the image reconstructed by the TOMP algorithm using block method are better than those reconstructed by directly using the TOMP algorithm.From the principle analysis,this is because the block method uses the spatial correlation in the plane of image,so it can effectively represent the original image with much less data than the original image.