Study of fusion of visible and IR images based on their correlation

Two developed color image fusion techniques, the MIT scheme and the TNO scheme, are applied to visible and infrared images. They are enhancement and nonlinear methods. This paper provides a discussion of the linear fusion scheme once again. At least two conclusions can be testified by the fusion experiment on two sensor images in this paper: one is the linear fusion scheme has good performance in some images modalities; the other is the linear fusion has better performance in the fused color space YUV than in the RGB. A key problem of the linear fusion, how to assign or weight sensor images to three channels of a color space, was studied with the correlation between dual-sensor images, and experimental results show that the assignment or weight is related with the correlation between sensor images. But this study is mainly for the Y channel in the interest of preserving the details from both of two sensor images, and the relationship between the assignment or weight of a linear fusion and the correlation between sensor images need to be studies further in future.