Technology Digital Infrared And Visible Image Fusion

In order to solve collaborative multi-source image sensor data, Multi-source image fusion techniques is developed, its purpose is to optimize and select the multi-source image data based on established strategies, abandon the redundancy information as much as possible, while trying to retain effective information, to obtain a fused image of multiple images. Multi-source image fusion technology has been had some successful applications in both military use and civil use, and has broad prospects for development. However, current image fusion algorithms have flaws such as detail information losses or target is not sharp.To solve the problems above, the main research works of this paper are: introduce the development and application of domestic and international image fusion techniques, review the image fusion algorithms and point out the weakness of these algorithms. Use the coaxial structure in the visible and infrared optical image fusion system to increase the accuracy of the image registration and decrease the time consuming. Preprocess to the collected infrared and visible images. Propose an image fusion algorithm based on the non-subsampled contourlet and image region segmentation, which will use the fuzzy C-means cluster image segmentation algorithm to divide the infrared image into two parts:the background and the target, and use different coefficients fusion strategy to fuse the high frequency and low frequency of them, which can effectively highlight interest information based on regional characteristics and achieve better fusion results. Colorize the fused image of the proposed algorithm in YUV channels, so that the colorized image contains a wealth of information and the colorized fused image fits for the human visual characteristics.Verify the validity of the proposed algorithm by comparing it with some other classical fusion methods. Experiments show that the fused image by the proposed algorithm highlight the target and preserve the details well, also the proposed algorithm has broad applicability.