Research On Heterogeneous Image Registration And Fusion Methods Based On Salient Contour

With the improvement of sensor technology,the image data obtained by multiple sensors presents diversity and complementarity,which realizes the multi-modal representation of the target scene.To comprehensively use heterogeneous images to achieve more accurate and comprehensive expression of the target field become a research hotspot in the field of image processing.Heterogeneous image registration and fusion technology are important parts of image processing technology.The registration method determines the effect of subsequent image fusion processing,and the fusion method realizes the efficient expression of source information.These two technologies complement each other,which has theoretical support and practical significance for improving the ability of target detection and prcision guidance.Synthetic Aperture Radar(SAR)has been applied to generate two-dimensional(2D)high-resolution images day and night under all weather conditions at a long distance.SAR images can clearly represent the contours of ground objects and have rich texture features.The visible light sensor has the advantage of fast imaging,and the optical images can intuitively reflect the regional attributes due to the rich spectral information.Therefore,the SAR image and optical image registration and fusion methods are deeply studied in this paper.The texture features of SAR image and spectral features of optical image are fully used to enhance the information of the target scene,so as to improve the ability of remote sensing image information mining and remote sensing image interpretation.The main tasks completed in this paper are as follows:1.Summarize and analyze the imaging mechanism and image characteristics of SAR image and optical image.Then,the basic framework,basic principles and basic classification algorithms of image registration and image fusion systems are studied respectively.Finally,the visual evaluation and objective evaluation indicators of image registration and image fusion systems are analyzed respectively.2.To solve the registration problem of large differences in image grayscale and texture caused by the difference imaging mechanisms of SAR images and optical images,and the traditional image registration methods cannot achieve heterogeneous image registration.This paper proposes a heterogeneous image registration method based on local contour points.Firstly,the enhanced Gamma?MAP filter is used to suppress speckle noise,and the CLAHE method is used to enhance the SAR image to improve the contrast and highlight image feature.Then,for salient contour features in heterogeneous images,Harris operator is used to detect stable invariant points in contour features,the arc length and chord length ratio feature is designed to complete the coarse registration,and the contour tree descriptor is designed on the basis of the coarse registration to complete the fine registration.The feasibility and robustness of the proposed method are verified by error analysis and simulation experiments.3.To solve the problems of missing edge texter details and spectral distortion in SAR image and optical image fusion results.This paper proposes a multi-scale heterogeneous image fusion method based on latent low rank representation.Firstly,the optical image is transformed by HIS transformation,the brightness components of optical image and SAR image are transformed by NSCT to obtain different subband components.Then,the low-frequency coefficient is represented by latent low rank to obtain low rank image and significant image,and different fusion rules are used to obtain low-frequency fusion results.For the high-frequency coefficients,the fusion method is the combination of the new sum of modified Laplacian and region gradient.Finally,the fusion results are obtained by using the inverse NSCT transform and the inverse HIS transform.The simulation results demonstrate that the proposed method has better performance.