Infrared And Visible Image Registration And Fusion Technology

With the rapid development of imaging technology and digital image processing, the applications of multi-sensors become feasible and effective. And many problems in military and civilian life have been successfully solved by the combination of infrared and visible images.The multi-source images should be aligned before using them. The basic theory and models of image registration are described first, as well as many effective matching measures proposed in current literatures. The discussion of the infrared and visible image registration is highlighted. According to the characteristic of the two image types, a new registration algorithm is proposed based on the corner points of the edges. The new corner detection method provides more corresponding points than other ways. The points are well matched via the comprehensive measure including curve moments and the gradient mutual information by analyzing the structure similarity.The multi-source image fusion is an effective way of information utilization. After the image registration, the basic theory and the classification of image fusion are summarized, as well as the most basic fusion rules and the image quality index. The fusion algorithms of this paper are established on the multiresolution image fusion using wavelets, and the aim is to find an effective rule for the infrared and visible image fusion. An improved Petrovic fusion rules and a contrast enhancing method are introduced and both of them consititute the paper's fusion algorithm. The algorithm is proved to be more advantages over several other methods by experiments. The contrast enhancing method is based on the analysis of infrared and visible light image characteristics and the intention for objectives. And the experiments of applying the contrast enhancing method to other fusion rules prove that it can effectively improve the quality of the images.