The Infrared and Visible Light Image Fusion Based on the Non-subsample Shearlet Transform and Heat Source Concentration Ratio

Abstract

Image fusion is the technology that use image processing algorithm to integrate different and complementary information form two or more images to create a new collection which is more accurate and convenient for the further use. As one significant component of the Image Processing, Image fusion is widely employed in many aspects, such as medical diagnose, military reconnaissance and remote sensing survey. To get a fused image which combines the target information and features of infrared and visible light image, a fusion method based on the Non-subsample Shearlet Transform (NSST) and heat source concentration ratio is presented in this paper. Compared with the traditional contourlet transform, NSST can overcome the limitation in directional decomposition and has excellent shift invariance. The input is decomposed to two parts by the transform, a new fusion rule which adopts the heat source concentration ratio and space frequency to retain as more important information as possible is presented in the low-frequency part and information entropy serves as the measurement in the high-frequent part. Simulation shows that the new fusion scheme can obviously improve the quality of fusion image and makes the heat source information prominent.