¡¡	Chinese Journal of Computers   Full Text
Title	Fusion Method of Multi-Spectral Image and Panchromatic Image Based on Four Channels Non-Separable Additive Wavelets
Authors	LIU Bin1) PENG Jia-Xiong2)
Address	1)(School of Mathematics and Computer Science, Hubei University, Wuhan 430062) 2)(Institute of Image Recognition and Artificial Intelligence, Huazhong University of Science and Technology, Wuhan 430074)
Year	2009
Issue	No.2(350¡ª356)
Abstract &   Background	Abstract In order to solve the problem that separable wavelets are not symmetry and can not get high spatial resolution image. This paper presents a construction method of a class of wavelets, non-separable wavelets which dilation matrix is £Û2,0;0,2£Ý and has compactly support, symmetry, orthogormality and its application in the fusion of multi-spectral image and panchromatic image. Using matrix expansion, a method constructing four channels 6¡Á6 filter bank is presented. Quickbird images are decomposed and reconstructed through additive wavelet mode using the low-pass filter of the filter bank. Three fusion modes called NAWS, NAWRGB and NAWL are proposed in the fusion of multi-spectral image and panchromatic image. The fusion performance of the fusion method is evaluated using objective index. The experiment results show that this method have good visual effective in the preservation of spectral quality and spatial resolution information. This fusion method outperforms tensor product wavelet fusion method, separable additive wavelet fusion method and two channels non-separable additive wavelet fusion method in the preservation of spatial resolution information and can get more information from the source images. The fusion method can also preserve good spectral quality. Keywords non-separable wavelets; filter bank; image fusion; multi-spectral image; panchromatic image Background Image fusion is the hot research field of image processing in recent years. Separable wavelets have been successfully applied in the image fusion, but the construction of non-separable wavelets has not been solved perfectly. Its applications in image fusion were begun a few years ago. The research group has constructed several kinds of non-separable wavelets filter banks and used them in the image fusion such as multi-focus images, visible image and infrared image, remote sensing image. Jorge N¨²¢‹ez et al had proposed a fusion method of multi-spectral image and panchromatic image based on separable additive wavelet and ¢~ trous algorithm. This paper will extend the fusion method of multi-spectral image and panchromatic image based on separable wavelet to non-separable additive wavelets, and provide a class of fusion method of multi-spectral image and panchromatic image based on two-dimensional four channels non-separable additive wavelets. This work was supported in part by the National Natural Science Foundation of China (No.10477007), the Natural Science Foundation of Hubei Province under project (No.2006ABA015), and the important foundation of Hubei Provincial Department of Education under project (No.D200510004). These projects aim at construction methods of non-separable wavelets and its filter banks. Its applications in image fusion and computer vision (such as airplane automatic landfall) are also main research focuses. The research interests of the group include wavelet theory and applications, image fusion and registration, image recognition, computer vision, motion detection and tracking, intelligent automation. The group has proposed several kinds of image fusion methods such as image fusion method based on symmetrical non-separable wavelets filter banks, image fusion method based on dissymmetry non-separable wavelets, image fusion based on two channels non-separable wavelets, image fusion method based on two-dimensional two channels non-separable additive wavelets. And a new fusion method of multi-spectral image and panchromatic image based on four channels non-separable additive wavelet is presented in this paper.