Research on Medical Image Fusion Based on Orthogonal Wavelet Packets Transformation Combined with 2v-SVM

This paper presents a novel medical image fusion algorithm which imports 2v-SVM -an adaptive SVM learning algorithm to medical image fusion. We combine it with orthogonal wavelet packets to generate a new image fusion rule, which intelligently constructs the "good and bad features-classifier" for improving image fusion. Then we construct a new sort of linear weighted fusion arithmetic operator, which retains the remarkable features and eliminates the redundant information of the two images. We respectively fuse two pairs of images: CT> MRI images and MRI, PET images effectively and obtain good visual effects and quality of fused image. The experiments prove that the proposed algorithm resolve the deficiencies of existing algorithms in characteristics representation and acceptance or rejection of information and provide an advanced method to medical image fusion.