Fusion Algorithm Of SAR And Visible Images Based On Salient Features Decision Rules

Image fusion has a very important position in the field of digital image processing,and its research background is very broad.The image fusion processing can extract two or more source images according to the required requirements and extract their respective characteristics according to certain rules.The obtained fusion image results need to be combined with the respective advantages and complementary information in the image,which is compared with any single before-fused images have the advantage of being more precise,rich and reliable,and on the other hand,this also compensates for the limitations of each type of sensor itself.The fusion of SAR and visible light images studied in this paper is mainly to extract the complementary feature information in SAR image and visible light image,and merge them into a fused image with a certain feature through certain rules.Comparing some traditional fusion algorithms,they will have some contradiction between spectrum and detail,and will also be affected by image noise and a lot of details lose after denoising.The main research work and research areas of this paper include the following aspects: learning and analysis of the domestic and international progress and image meaning of image fusion;analyzing and comparing SAR images and visible images from physical properties,imaging mechanisms and applications;enumeration and research According to the traditional algorithm of image fusion,based on the texture characteristics of SAR image and the spectral characteristics of color visible light image,and referring to the advantages of wavelet transform,the fusion algorithm of traditional energy amplification is studied and improved from multi-resolution angle,and the target extraction and noise are proposed.The method of removal;through the analysis and simulation research,a fusion algorithm based on salient feature regions is proposed.The spatial region of SAR image is divided into bright target region,dark target region and non-target region according to the feature information of the image.For different feature regions,using different strategies for target retention,noise reduction,and spectral retention.The target area(bright target and dark target area)adopts the feature information retention feature,and the non-target area adopts the reservation fusion to weaken the noise intensity value while retaining the background color of the target-less area to ensure that the dark target is not Lost.For the practical effect of image fusion,this paper uses objective and fair evaluation indicators to evaluate the fusion results.However,the traditional structural similarity(SSIM)index is based on the evaluation index of image gray scale,which cannot reflect the advantages and disadvantages of image fusion in the fusion of noisy images.In order to solve this problem,this paper proposes structure retention(FSIM),which is different from SSIM.FSIM throws away the statistics based on gray pixel information,and studies the edge structure and direction gradient information,and then ignores the influence of noise on the target information.The algorithm for preserving detail edges and removing noise in the image of the fusion result has a good evaluation effect.In this paper,spectral distortion(D)and structure retention(FSIM)are used as the main evaluation indexes of SAR image and color visible image fusion in spectral preservation and structure preservation.Mutual information(MI)is used as an auxiliary reference index for fusion evaluation.From the simulation results,it can be concluded that from the subjective analysis,the fusion image color information is excellent,the detail information is smoothly integrated,and objectively analyzed,all kinds of evaluation parameters are at a superior level,which is for further image processing(image segmentation,image Feature extraction,target recognition,etc.)is of great help.A comprehensive comparison of representative wavelet replacement fusion,IHS-DWT fusion algorithm and wavelet sub-regional algorithm,the proposed method has obvious advantages in subjective visual representation and objective evaluation parameters.