Fusion of Infrared and Visible Images Based on Pulse Coupled Neural Network and Nonsubsampled Contourlet Transform

Abstract

This paper presents a new fusion algorithm which can effectively solve the problem that has unobvious infrared target and low contrast in infrared and visible image fusion.This paper's innovation point is its fusion rule.Other algorithms' fusion rules usually use pulse coupled neural network (PCNN) and region characteristics to select low frequency or bandpass subband coefficients.The proposed algorithm innovatively applies improved PCNN and region characteristics to the selection of both low frequency and bandpass subband coefficients in nonsubsampled contourlet transform (NSCT) domain.First, the subband coefficients of original image are obtained by NSCT.Then, the decomposed subband coefficients are processed by using PCNN, whose fire mapping images are obtained.The method of region standard deviation is used to choose the fusion coefficients of fire mapping image, which acquires more image information in low frequency part.For bandpass subband coefficients' fire mapping image, the method based on region energy is adopted for fusion coefficients, which makes the bandpass part capture more energy.Finally, the fused image can be obtained by inverse transform of NSCT.Compared with typical wavelet-based, NSCT-based and NSCT-PCNN based fusion algorithms, experiment shows that the new proposed algorithm can improve the fused image's objective evaluation index significantly, obtaining a prominent infrared target and better fusion image quality.