Compressed Sensing Remote Sensing Image Reconstruction Based on Wavelet Tree and Nonlocal Total Variation

Compressed sensing reconstructs data using much less sampling data, generally as O(K+Klogn), compared to Nyquist theory. K is the sparsity of data and n is the length of data. Furthermore, according to the basic principle of structured sparsity theory, for a standard K-sparse data, the measurements need for reconstructing original data can be further reduced from O(K+Klogn) to O(K+logn). The wavelet coefficients have tree structure and can be used in compressed sensing. The nonlocal total variance (NLTV) is highly effective in sharping image edges and preserving fine details. It also performs well in getting rid of the block effects caused by total variance (TV). Consequently, a new model based on NLTV and wavelet tree is proposed in this paper for solving the compressed sensing remote sensing image reconstruction. Experiments show the well performance of our model in reconstruction accuracy compared to other methods.