Compressed Sensing Based on the Integer Wavelet Transform for Image Processing

Abstract

To improve the quality of a reconstructed image, a compressed sensing algorithm based on the integer wavelet transform was proposed, Compared with traditional wavelet transform, the integer wavelet transform reduces the error made by the floating numbers, and remains the multi-resolution property, which improves the quality of the reconstructed image, According to the well Sparsity of high-frequency coefficients, only measured high-frequency coefficients. Different measurement matrices were compared. Orthogonal Matching Pursuit algorithm was used while reconstructing. Compared with the original wavelet compressed sensing algorithms, the quality of the reconstructed image can be greatly improved with less measurement numbers. For the same measurement numbers, PSNR of the proposed algorithm can be improved about 2-6dB.