Research On Image Fusion Algorithm Based On Region And Multi-scale

In recent years, because of its important role in machine vision, image analysis and other fields, the technology of image fusion attracts more and more attention. Compared with a single source image, the fused image using image fusion technology can be more comprehensive, more detailed and more reliable. In this thesis, by analyzing the imaging characteristics of infrared and visible images, a new image fusion algorithm based on regional window and multi-scale has been proposed, which used PCNN segmentation method. Then, by comparing the simulation results of different algorithms, combining with the subjective visual and objective parameters evaluation method, the effectiveness of the proposed algorithm in this thesis has been verified. The main work of this thesis is as follows:(1) The thesis analyzed the imaging characteristics of the infrared and visible images and the transformation principle of the multi-scale analysis tool, and mainly introduced the translation invariance of NSCT, and researched the influence of the multi-scale tool’s shift-variant to the fusion algorithm through simulating and comparing.(2) The thesis proposed a new image fusion algorithm based on regional window and multi-scale, which used PCNN segmentation method. By compared with other related algorithms, the superiority of the fusion rule and rationality of the segmentation method used in the algorithm have been verified.(3) The thesis analyzed several groups of infrared and visible images in different application scenarios, by using the image fusion algorithm proposed in this thesis. Combining with the subjective visual and objective parameters evaluation method, the applicability of the proposed algorithm has been verified.(4) Based on the GUI module of Matlab, this thesis has implemented the image fusion simulation software. Thus, the visualization of the algorithm and the comparion and analysis between algorithms became more convenient.Through the research above, the image fusion algorithm based on regional window and multi-scale proposed in this thesis was used to fuse the infrared and visible images in different application scenarios. The fusion result shows that the target and background information of the source images has been extracted and preserved relatively good. Therefore, the algorithm is suitable for the fuse of the infrared and visible images.