Improved Algorithms for Compressed Sensing Based on the Multi-scale Wavelet Transform

The designing of sensing matrix has been our focus in the research of the Compressed Sensing (CS). In this paper, according to the properties of wavelet transform sub-bans, we improved the sensing matrix of the classical algorithms for compressed sensing based on the multi-scale wavelet transform. By using the weight matrix, the more information of the wavelet coefficients which is more important for the signal reconstruction can be reserved after measurement, and the PSNR of the improved algorithm was improved about 0.5~0.8dB.