Study On The Methods Of Image Fusion Based On Multi-resolution Analysis

Image fusion is a processing which blends the complementary and common features of a set of images to generate a new image.The blending image has ‘better' information from the view of subjective and objective analysis.The multi-resolution analysis is an important idea of image processing.that is the strategy of ‘divided and conquer' method.As we all know,a nature image contains a lot of information including the texture,corner,contour and the smoothing area and so on.In short,including the rough and the detail information.Using a Multi-resolution method to decompose images,we can get a set of sub-images which include the details of the source images.Using different fusion strategies to fuse different parts and we can get an optimal result.The main work of this paper is around the multi-resolution analysis research in the development of image fusion algorithm,and fusion for multi focus image fusion algorithm is proposed,based on the combination of focusing region detection and non-subsampled contourlet transform.The main content of this paper can be summarized as following three parts:1.Research and analysis of various objective evaluation criteria for image fusion.The objective and subjective evaluations of image fusion results is an integral part of the fusion algorithm.Effective fusion quality evaluation criteria can be used to verify the effectiveness and superiority of the fusion algorithm.In this paper,we mainly study the evaluation criteria of image signal to noise ratio,cross entropy,information entropy,mutual information and edge gradient operator.The experimental results show that these indexes are correct and effective.2.Research on the theory of multiple resolution analysis for image fusion.In view of the image fusion algorithm,the principle and advantages and disadvantages of various multi-resolution analysis methods are analyzed and studied.Multi resolution image fusion based on contour wave transform is studied in the paper.Through pyramid transform,an image can be decomposed into a set of resolution decreased sub images,then through calculating the residual Pyramid get the image details and approximations,for the image in different resolution provides the condition of convergence.However,there is a large amount of redundant information among the sub-images,and the sub images are not directional.These problems limit its application in image fusion algorithm.In view of these problems,the image fusion algorithm based on wavelet multi-resolution decomposition is studied.Wavelet transform can achieve the time-frequency localization of the image analysis,image spectrum of orthogonal decomposition.The image signal energy is transferred to a small number of coefficients to achieve a sparse representation of the image.And the decomposition of the sub image with three directional information.Overcome the shortcomings of the Pyramid transform.Image fusion needs to preserve and accurately describe the important information in image fusion.The two-dimensional wavelet is from one-dimensional wavelet tensor form.It is isotropic,and can not be sparse and accurate representation of the edge profile information.Compared with the wavelet transform,the shape of the rectangular support domain can meet the requirements of the wavelet transform.Due to the direction of the contourlet filter structure,making it with more directional selectivity,the image fusion effect was enhanced.3.For the fusion of multi-focus images,a multi-focus image fusion algorithm based on non-subsampled contourlet transform(NSCT)and focused region detection is proposed.For multi-focus image fusion,the detection and extraction of the focused region is the key to the fusion.Because the focused region also belongs to the significant region,the GBVS saliency region detection algorithm is used to detect the focused region.The saliency map is obtained by using the spatial frequency method to get the focused region boundary.Then the watershed algorithm and morphological processing method are applied to integrate the small areas,and to repair the fracture edges and remove the pseudo-focused region.The focused areas with clear edges are fused directly,and the defocused regions are fused with the NSCT fusion method.The proposed algorithm is compared with the wavelet based multi-focused image fusion algorithm and the multi-focused image fusion algorithm based on the directly NSCT transform.Theoretical and experimental results show that the proposed algorithm has better performance under the subjective and objective evaluation.