Image Compression Based On Morphological Component Analysis And Compressive Sensing Theory

Efficient storage and transmission of information is an important link in the field of information and communication. The mass of data in practice has greatly restricted the storage and transmission efficiency of digital signal. It is an effective way to improve the storage and transmission efficiency of digital signal. In recent years, some digital image compression method based on compressed sensing and sparse representation theory can be used to achieve high quality digital image compression and reconstruction based on Nyquist sampling theorem. The new method is an effective method to improve the storage and transmission efficiency of digital image. It has a wide application prospect in medical imaging, pattern recognition, wireless communication and radar remote sensing. Based on the theory of sparse representation of signals, compressed sensing and morphological components, the paper studies the digital image compression technology. The main work is reflected in the following three aspects:1. The theory of sparse representation, compressed sensing and morphological component decomposition is studied. Learning and numerical implementation of the JPEG and JPEG200 two kinds of classical digital image compression method, as well as the compression method based on wavelet transform and single layer wavelet transform, and the simulation experiment, the experimental results are analyzed.2. learning and studying the Contourlet transform. The gray level and color image compression of Contourlet sparse representation combined with compressed sensing theory is realized, and the experimental results are analyzed.3. in the image compression method based on the theory of compressed sensing, the introduction of morphological component decomposition process, proposed an improved morphological component decomposition combined with compressed sensing theory of gray and color image compression method and numerical implementation. Firstly, the structure and texture components of the image are captured by RDWT and WAT redundant dictionary. Secondly, the structure and texture components are compressed sensing measurement. Experimental results show that the new method can reconstruct higher quality images in the case of less data storage.