The Multi-source Image Fusion Techonology By Muliti-scale

Now the ability to obtain images through sensors is improving.But sometimes the single sensor can not meet the needs of the target detection,we need a new technology to process the image that collect by multiple sensors.Multi source images fusion is a new techonologyto solve that.It is according to some algorithm to process the image that collects by multiple sensors,in order to get a more complete and accurateimage.This article is to research this technology,the imagesare polarized images andvisible images.Polarization image can obtain polarization information about the target,it can enhance contrast between object and background in image according to the polarizationinformation,increase the clarity of the target in the image.Because polarization image can reflect more information about object surface material and texture,so fuse it with visible image to achieve the goal of target identification become possible.In this paper,the first step is to calculate the transformational matrix between the polarization image and visible image by the improved sift algorithm.The second step is to fuse two images by the contourlet transform.Because the contourlet transform has the characteristics of multi-directional and anisotropy,it makes up for the lack of multi directional features of wavelet transform,so we choose this method to fuse theimage.The experiment images are the rose imagescollected in school laboratory,real roses and artificial roses are stochastic allocation.It can distinguish the roses easily from the experimental results.The results show that the algorithm can well fuse two kinds of source image information,thus it can achieve the simple target recognition.