An Image Fusion Algorithm based on Nonsampled Contourlet Transform

Abstract

In this paper, a multisensor image fusion method based on non-sampling Contourlet transform is proposed. The transform has translation invariance, is suitable for expressing images with rich detail information and direction information, and can avoid the ringing effect introduced by general methods to the fused images. The image is decomposed by the transform to obtain high and low frequency components. The high-frequency component uses PCNN as the fusion rule, and the low-frequency component uses the golden section method to search the optimal low-frequency fusion weight, so as to fuse the low-frequency subband coefficients of multi-sensor images adaptively. Comparative experimental results show that the proposed method can achieve better fusion effect.