Block Compressed Sensing Observation Matrix Optimization Algorithm Based on Block Target

Block Compressed Sensing Observation Matrix Optimization Algorithm Based on Block Target

In order to eliminate the blocking effect of block compressed sensing algorithm, an optimization algorithm of block compressed sensing observation matrix based on block target is studied. The theory of block compressed sensing is to segment the original image with fixed size to obtain the sub blocks, arrange the texture of each sub block, use the compressed sensing observation matrix to sample each sub block, and optimize the observation matrix of block compressed sensing to eliminate the blocking effect. The block target method uses sparse orthogonal basis to make the processed block target conform to sparsity and orthogonality. The sparse coefficient vector is obtained by the basis inverse transformation, and the reflection coefficient of block target is obtained by using the sparse coefficient vector to optimize the observation matrix of block compressed sensing. The experimental results show that the peak signal-to-noise ratio (PSNR) is higher than 31dB when the algorithm is applied to image reconstruction, and the relative support set error of different sparsity and observation times is low, which can effectively eliminate the blocking effect of block compressed sensing algorithm.