Fusion Of SAR And Infrared Image For Information Enhancement

Infrared and Synthetic Aperture Radar(SAR)image fusion is an essential branch in the field of image information enhancement.The sensor imaging principles of infrared images and SAR images are quite different: infrared imaging technology extracts the target through the difference in thermal radiation between the target and the background,and can penetrate a certain thickness of soil layer or even concrete layer,but infrared imaging technology is easily affected by clouds,rain and fog and other environmental factors and SAR imaging technology is not affected by weather,time and other factors to achieve long-distance detection,easy to detect objects with high reflectivity such as metals.Therefore,the content and imaging conditions of these two imaging methods have complementary characteristics.If a certain fusion algorithm is used to fuse the infrared and SAR images,the fused image can contain more texture detail information,improve the visual recognition of the image,and enrich the information of key targets,laying the foundation for further target detection.Aiming at the problems of poor image quality,unclear edge contour,low efficiency,poor visibility,and low target detection efficiency of existing fusion algorithms,this thesis proposes two fusion algorithms for infrared and SAR images.The main research contents of this thesis are as follows:(1)Aiming at the weak target in a single image,this thesis proposes an infrared and SAR image fusion algorithm based on non-subsampled contourlet transform(NSCT).The preprocessed infrared and SAR images are firstly decomposed by NSCT to obtain lowfrequency and band-pass orientation images.The low-frequency image contains most of the information and energy of the original image.First,the local window weighted average energy is used to smooth the low-frequency image to obtain a low-frequency approximate image.Then,the sum of modified Laplacian is used to extract the edge detail information of the low-frequency image to obtain The low-frequency edge image;Finally,the corresponding pixels of the low-frequency approximate image and the low-frequency edge image are multiplied to obtain the feature map,which is subsequently normalized to determine the weights of the low-frequency infrared image and SAR,and realize the lowfrequency image fusion.For the bandpass direction image,firstly extracting the edge contour,contrast and brightness information of the image through phase consistency,local sharpness and local energy,then exponentially multiply them to obtain the feature map,and finally make a decision on the feature map to obtain the bandpass direction fused images.In order to test the superiority of the performance of the algorithm proposed in this thesis,two sets of infrared and SAR images are used for simulation experiments.Compared with other image fusion algorithms,the image fusion algorithm in this thesis obtains the best fusion image quality.At the same time,the fusion image is used for target detection,and it is concluded that the fusion algorithm can improve the accuracy of target detection.(2)Aiming at the situation that the edge contour information of a single image is blurred,this thesis proposes an infrared and SAR image fusion algorithm based on non-subsampled shearlet transformation(NSST).The SAR image contains partial coherent speckle noise,so the bilateral filtering algorithm is used to filter it firstly.Then,the preprocessed infrared and SAR images are decomposed by NSST to obtain low-frequency and band-pass images.Finally,the gradient fusion method is used for both the low-frequency and band-pass images.The low-frequency image uses Sobel algorithm to get its gradient.However,since the bandpass direction image contains more texture details,the Laplacian operator with stronger edge positioning ability and better sharpening effect is used to obtain the gradient of bandpass direction image.In order to test the superiority of the performance of the algorithm proposed in this thesis,three sets of infrared and SAR images are selected for simulation experiments.Compared with other image fusion algorithms,the image fusion algorithm in this chapter can greatly improve the edge definition of the fused image,which is in line with the human visual perception.