Research On Infrared And Visible Image Fusion Algorithm

As an important branch in the field of image fusion,infrared and visible images fusion has been widely used in engineering technology and scientific research.Infrared and visible images come from different imaging sensors,and images with different characteristics are collected for the same scene.Infrared and visible image fusion technology not only solves the problem of fuzzy background and missing details of a single infrared sensor,but also overcomes the dependence of visible light sensors on environmental conditions,and constructs images with thermal radiation sources and multi-spectral information.The fused image is more in line with human vision and convenient for subsequent research,so it has more practical value.This thesis mainly studies high performance infrared and visible image fusion algorithm,the main research contents are as follows :1.Multi-scale transform is a recognized and efficient decomposition tool.Image fusion based on multi-scale transform is an active research direction in the field of infrared and visible image fusion.In this thesis,the most advanced Non-Subsampled Shearlet Transform(NSST)was selected as a multi-scale transform tool.The source images were decomposed into sub-images with different resolutions to extract different features,different fusion rules were used to fuse the sub-images,the final fusion image was obtained by inverse NSST.The experimental results showed that compared with other multi-scale transform methods,the NSST-based image fusion algorithm could extract more image features and have more effective expression of the scene.2.Aiming at the problems that the infrared target is not prominent and the detail information of visible image is lost in the existing fusion algorithm results,this thesis proposed a fusion algorithm based on NSST decomposition and nonlinear enhancement.The source images were decomposed by NSST to obtain the corresponding low-frequency and high-frequency images.On the basis of infrared low-frequency images,the ' S ' nonlinear enhancement function with adjustable parameters was formulated to obtain the fusion weight of infrared and visible low-frequency images.The high-frequency images were fused by the method of ' absolute value-max '.Finally,the final fused image was obtained by inverse NSST.The experimental results showed that the proposed algorithm could enhance the contrast between the target and the image while preserving a large number of visible image details..3.Aiming at the color difference problem in color images fusion,this thesis proposed an infrared and color image fusion algorithm based on HSI and NSST.The I component of color image was obtained by HSI transform,and it was decomposed by NSST with infrared image at the same time.The decomposed low-frequency images were fused by regional energy and intuitionistic fuzzy set.The decomposed high-frequency images were fused by the fusion rule based on wavelet transform.The new I component was obtained by NSST inverse transform.Finally,the fused color image was obtained by HSI inverse transform.The experimental results showed that the proposed algorithm had a high degree of restoration to the color of the color image background,and the fusion results obtained were more suitable with the human visual characteristics.