Jindong Xu
Yantai University School of Computer and Control Engineering
China
Title: Multi-scale sparse decomposition of image and its application in Remote Sensing image fusion
Biography
Biography: Jindong Xu
Abstract
Morphological component analysis (MCA) is a successful example of a sparse image decomposition algorithm. Building on MCA, a multi-scale sparse image decomposition method, called m-MCA, is presented in this paper. M-MCA combines Curvelet Transform bases and Local Discrete Cosine Transform bases to form the decomposition dictionary and controls the entries of the dictionary to decompose the image into texture component and cartoon component. From the aspect of the amount of information, a remote sensing image (RSI) fusion method based on multi-scale sparse decomposition is proposed. Via sparse decomposition, the effective scale texture component of high resolution RSI and cartoon component of multi-spectral RSI are selected to be fused together. Compared to the classical fusion method, the proposed fusion method gets higher spatial resolution and lower spectral distortion with a little computation load. Compared to sparse reconstruction fusion method, it achieves a higher algorithm speed and a better fusion result.