Image reconstruction in Compressed Sensing based on single-level DWT

摘要

Wavelet transform is commonly used in Compressed Sensing. To improve the image reconstruction quality in CS based on single-level DWT, two improved methods are proposed. Row-column average method processes the row vectors and column vectors of the coefficient matrices separately to get two reconstructed images, and then takes the average of them as the final result. Threshold method sets a threshold to improve the sparsity of row vectors or column vectors of the coefficient matrices. Experimental results show that the two methods can significantly improve the quality of reconstructed images.