Multi-focus Image Fusion Based on Multiple CNNs in NSCT Domain

Abstract

In order to overcome the boundary information loss in the image fusion with single convolutional neural network, this paper proposes a novel multi-focus image fusion with multiple convolutional neural networks in nonsubsampled contourlet transform (NSCT) domain. First, the source images are decomposed into a low frequency sub-band and a serious of high frequency sub-bands by using NSCT. Second, a corresponding CNN model for each level of high frequency sub-bands is trained to fuse them. Then, an averaging rule is employed to fuse the low frequency sub-bands. Finally, the fused image is reconstructed by performing inverse NSCT on the fused sub-bands. Experimental results illustrate that the proposed method is superior to several existing multi-focus image fusion methods in terms of both executive evaluation and objective evaluation.