Medical image fusion based on GPU accelerated nonsubsampled shearlet transform and 2D principal component analysis

摘要

This paper presents a new medical image fusion method based on the GPU accelerated nonsubsampled shearlet transform and 2D principal component analysis. In order to keep the smoothness of pixels distribution in fused image, the subbands coefficients after transform are divided into blocks. A global and regional combined fusion strategy is used to reduce the distortion exists in coefficients changes. The experiment results show that the proposed method could provide an improvement on the quality of fused image when comparing to other multiscale transform based methods.