Research Of Image Fusion Algorithm Based On Nonsubsampled Contourlet Transform

Along with the increase in image information acquisition means, we getimages of different types of information exploding. Between these data, both arecomplement and redundant, how to extract the more useful, more refined andhigher quality data from these both complementary and redundancy of the imagefor human decision or decision reference for the decision system of artificialintelligence, which has become an urgent problegm need to solve, so the imagefusion technology have become a research hotspot for many scholars.Imagefusion is a technology that combine the advantages of two or more imagesorganically according to certain rules and fusion algorithm to produce a newimage.The new image not only makes up for the deficiency of the single imagegets more abundant and precise information than single image,but also makes itclearer in visual quality and much easier to understand, the application scope oftheir respective information is greatly expanded.This paper introduces in detail how to solve the practical problem on theapplication of nonsubsampled Contourlet transform to the multi-focus imagefusion and image fusion in the problems.Firstly, this paper introduces the conceptof image fusion, the purpose and significance of the research, the classification ofimage fusion and common image fusion method.Secondly, this paper introducesthe basic theory of image wavelet transform. And it is discussed that thedeficiency and the advantages of applying wavelet transform in imageprocessing.Because of the shortage of wavelet transform, Contourlet transformtheory is introduced,but this theory has defects of mixing phenomenon,so wepropose the nonsubsampled Contourlet transform theory,which not only has theadvantage of more direction and multiscale,but also possess the characteristics oftranslational invariance compared with the Contourlet transform,and avoid thepseudo Gibbs phenomena in the Contourlet transform.Thirdly, The paper put emphasis on the application of non-subsampled Contourlet transform inmultifocus image fusion and medical image fusion, in multifocus image fusion,this paper studies a kind of fusion algorithm based on the combination ofimproved local entropy,directive contrast and local standard error, and alsostudies the selection of the optimal decomposition level and direction number ofnon-subsampled Contourlet transform,In medical image fusion,this paper studiesa fusion algorithm based on the improved local entropy and area energy,and thesimulation result is obtained by MATLAB simulation.Finally, the paper proposesseveral performance evaluation index of image fusion quality and compares thealgorithm in this paper with other fusion algorithms through subjective andobjective evaluation,and find that the image got by fusion algorithm in this paperis not only clearer on the visual effect,but also can display the image texturefeature and edge information is better.