DSA Image Fusion Based on Dynamic Fuzzy Logic and Curvelet Entropy

The curvelet transform as a multiscale transform has directional parameters occurs at all scales, locations, and orientations. It is superior to wavelet transform in image processing domain. This paper analyzes the characters of DSA medical image, and proposes a novel approach for DSA medical image fusion, which is using curvelet information entropy and dynamic fuzzy logic. Firstly, the image was decomposed by curvelet transform to obtain the different level information. Then the entropy from different level of DSA medical image was calculated, and a membership function based on dynamic fuzzy logic was constructed to adjust the weight for image subbands coefficients via entropy. At last an inverse curvelet transform was applied to reconstruct the image to synthesize one DSA medical image which could contain more integrated accurate detail information of blood vessels than any one of the individual source images. By compare, the efficiency of our method is better than weighted average, laplacian pyramid and traditional wavelet transform method.