Improved MP Spares Decomposition Method And It’s Application On Image Compression

In recent years, the process of science and technology has been accelerated constantly. As one of the most commonly used information carriers image has been widely used in various fields and then the digital image processing technology has gradually attracted people’s attention. As a basic part of digital image processing image compression has an important position in it. Matching pursuit sparse decomposition which has a simple decomposition result and has shown great potential in terms of image compression is a kind of non-orthogonal decomposition method. And in this paper it studied the decomposition method of matching pursuit under the background of image compression. The more specific content is as follows:First of all, it studied the basic theory of signal sparse decomposition. And then on the basis of the above research work, it came to calculation the process of MP sparse decomposition method, analyzed the complexity of the computational process, proved that matching error will exponential decline along with the decomposition process and illustrated the method of convergence. And in the simulation experiment, it illustrated the sparsity of decomposition results by using different number of atoms to reconstruct the digital signal in the Gabor atomic dictionary.Secondly, according to the characteristics of the image data, it studied asymmetric over-complete dictionary which is used for the image sparse decomposition and then came to discretization method of over-complete dictionary parameter which is generated when the image sparse decomposition is proceeding. On the basis of sufficient study of the artificial fish colony algorithm, it calculated every step’s inner product of the MP sparse decomposition method.Finally, aiming at the characteristics of image sparse decomposition method and artificial fish colony algorithm various, it proposed the further optimization method of foraging, gathering, following, and parameter setting. And then it proved that the image sparse decomposition method has achieved a good effect in image compression though the simulation experiment.