An adaptive sampling method of compressed sensing based on texture feature

An adaptive sampling method of block-divided compressed sensing for images based on textural feature is proposed. First, the SF (Spacial Frequency) is utilized to extract the textural features of image blocks. Then based on the textual features, each block is categorized into the smooth blocks or the textual blocks, and the basic sampling rate can be obtained simultaneously. To the textural blocks, we still use the basic sampling rate is modified adaptively by combing with the statistical characteristics of the coefficients in wavelet domain. To validate the effectiveness of proposed sampling method, the smooth projected Landweber (SPL) is employed to reconstruct the images, and the results are compared with other block-based compressed sensing (BCS) algorithms which proposed in recent years from the aspects of the objective index and the subjective visual impression. The experiment results show that when the compressing ratio is modest, the proposed method can improve the reconstruction quality of image evidently.