Thermal Infrared Image And Visible Light Image Fusion Technology Research

With the rapid development of infrared images in the military, civil and other fields, making fusion infrared thermal image with the visible image has been extensively studied and attention. Infrared image has a strong anti-jamming capability, can work around the clock, is good for covert target detection and other advantages, it is widely used in military, industrial and medical. But the infrared image has the disadvantages of low contrast, fuzzy background, poor edge detail representation and so on. And visible light images with abundant spectral information, a high degree of spatial discrimination, the details of the scene texture clear, so extraction and integration of useful information from infrared and visible images can be attain more accurate, clear and comprehensive information. Therefore, this paper proposes an adaptive image fusion algorithm based on non-subsampled contourlet transform(NSCT) and pulse coupled neural network(PCNN) is proposed, which is based on the analysis and summary of the relevant work of image fusion technology.NSCT transform effectively overcome the aliasing problems of the conventional multi-scale transform, it is not only has the multi-scale and multi direction characteristic, also has the translation invariance. So, after transformation a series of low frequency subband coefficients and high frequency directional subband coefficients are obtained. Then, according to the regional features of the sub bands formulate the corresponding fusion rules. PCNN has pulse release characteristics, capture characteristics, variable threshold characteristics, etc. It can be used in the development of fusion rules to obtain a good fusion effect. So, the low frequency subband coefficients are processed by the fusion rule based on the adaptive region energy, and the high frequency directional subband coefficients are selected based on the adaptive fusion rule of PCNN. Among them, this paper uses the improved spatial frequency as the link strength of PCNN, and carries on the consistency inspection to the ignition mapping chart, further improve the clarity and accuracy of image fusion.In this paper, several traditional image fusion algorithm and algorithm in this paper has been Matlab simulation, and using six objective evaluation indexes to evaluate the image fusion effect, such as average gradient, spatial frequency, standard deviation and so on. The results showed that: image fusion based on adaptive image fusion method proposed in this paper, not only has a good visual characteristics and rich edge details, and can get a higher evaluation index. Among them, the average gradient lifting 9.42%~34.662%, spatial frequency improved 12.114%~14.193%, standard deviation improved 28.661% ~ 30.032%.