A Novel Precise Decomposition Method for Infrared and Visible Image Fusion

Abstract

Decomposition is a necessary step for multi-modality source image fusion. Multi-scale transform (MST) is widely used in multi-modality image fusion. However, directly using MST to decompose source images to high and low frequency component for fusion is not precise enough. In this paper, we propose a precise decomposition method in non-subsampled contourlet transform (NSCT) domain. In our proposed method, NSCT is applied to source images decomposition for obtaining corresponding high frequency and low frequency sub-bands. The high frequency sub-bands of different decomposition layers carry different information. In order to obtain a more informative fused component of high frequency, maximum absolute value and Pulse Coupled Neural Network (PCNN) fusion rules are implemented for different sub-bands of high frequency components integration. An activity measure including phase congruency (PC), local measure of sharpness change (LSCM) and local energy (LE) is designed to enhance the detailed feature of low frequency fused image. The integrated high and low frequency components are then merged to a fused image. The experimental results show that the fused image obtained by this algorithm has good superiority in clarity, contrast, image information entropy and so on.