Fusion Of Night Visible And Infrared Images And Color Research

This paper mainly researches the fusion of night visible and infrared images. In order to meet human eyes, further, to research colorization methods. The fusion can realize that the description of the scene is more accurate and reliable. So, the color fusion is widely used in military field, security monitoring field. In this paper, specific work is as follows:Firstly, to research fusion algorithm, and analysis the advantages and disadvantages of each algorithm, from the perspective of multi-resolution analysis, so as to determine the best fusion scheme.Because NSCT(nonsubsampled contourlet transform) algorithm has many advantages, such as, Multi-scale, Multi-directional, Translation invariance,we employ NSCT to decompose source images, getting low and high frequency coefficients. Using the adaptive weighted fusion rule to fuse low frequency coefficients, And the fusion rule of high frequency coefficients is based on region variance.Secondly, to research false color and color transform based on color space, false color by mathematical model, establishes a mapping relationship from gray image and color image.Selecting a color image which is similar color statistics.In the color space, its three channels are less correlation, to achieve color transferring.Thirdly, to research PCNN, which is put forward based on animals' visual system.When PCNN is used to fuse images, which can preserve more edge features, for which human eyes are more sensitive. So, the fusion scheme, which combines NSCT with PCNN, is put forward.In detail, using NSCT to decompose source images, getting low and high frequency coefficients. For low coefficients,employing spatial frequency(SF) as the link strength. Both SF and low frequency coefficients are as the inputs of PCNN network.For high frequency coefficients, According to the values corresponding coefficients, to determine high frequency fusion coefficients.Finally, In order to improve the real-time for sequence images' colorization.Step1, to get false color fusion image.Step2,to choose natural color images as reference images.Step3,do color transform to achieve color correction. Step4, to establish color lookup table. Because thedistance is closer, the color is more similar. We can get the complete color table. By querying the color lookup table, to fast realize sequence images color fusion.