Research On SAR Image Registration Based On Feature Point

Synthetic aperture radar(SAR)has been widely used in various military and civil applications since it possesses the imaging characteristic of all-time,all-weather and high resolution.As the significant step toward SAR image interpretation,SAR image registration is the basic of SAR image fusion and change detection.During the last decades,it has become the research hot spot for foreign and domestic researchers.Focusing on the SAR images with different polarizations,bands,temporals and sensors,this dissertation studies the SAR image registration technique based on the studies of feature extraction and feature matching of SAR image.In this dissertation,several improved SAR image registration methods are proposed for providing strong noise robustness and high registration precision.The main contents and the acquired research results in this dissertation can be summarized as the following four parts:1.Considering the limitation of the SIFT method on SAR image registration,a novel SAR image registration method is proposed,which is based on the combination of SIFT,nonlinear diffusion,and phase congruency.In the proposed algorithm,the multiscale representation of an SAR image is generated by nonlinear diffusion,since it better preserves edges in the image as opposed to Gaussian smoothing,which is used in the original SIFT.Moreover,phase congruency information is utilized to remove the erroneous keypoints within the initial keypoints.Experimental results on real SAR images indicate that the proposed method is able to improve the match performance,which leads to an improvement on noise robustness and registration accuracy.2.Focusing on the limitation of the gradient-based feature descriptor and conventional Euclidean distance ratio matching criterion,a new point matching algorithm is proposed to align two SAR images.In the proposed method,by considering image patches as the basic units,a novel local descriptor including the intensity and geometric information is assigned to each feature point,which is more robust to speckle noise.Furthermore,a correspondence establishment scheme is introduced based on the reconstruction error ratio between feature points calculated by the sparse representation(SR)technique,which is designed for achieving accurate matches.Based on the obtained SR coefficients,a coordinate correction procedure is further proposed for improving the localization accuracy of the obtained correspondences.Both simulated deformed and real SAR images are utilized to evaluate the performance.Experimental results indicate that the proposed method yields a better registration performance in terms of both accuracy and robustness.3.In order to reduce the influence of speckle noise to feature extraction and improve the performance of feature descriptor with single-scale image,we propose a new image registration method for SAR image with multiscale image patch features,in which the sparse representation technique is exploited.Considering the influence of speckle noise on feature extraction,in the proposed method,a spatial correlation strategy based on stationary wavelet transform is adopted to select the reliable feature points from the initial keypoints in the reference image.By introducing multiscale image patch,a new feature descriptor is further designed to describe the attribute domain of feature points for higher discrimination.The corresponding points in the sensed image are established based on the minimum discrepancy criterion calculated by the sparse representation technique.Moreover,the local geometric consistency among a feature point and its nearest neighbor points is employed to remove the mismatches from the tentative matches.Experimental results show that the proposed method is competent to improve the registration performance substantially.4.In order to solve the problem that the conventional SAR image registration methods cannot deal with the registration of large SAR images,a non-rigid registration method is proposed to align two SAR images by registering two sets of feature points extracted from the images.In the proposed method,both point-wise background regional similarity and local spatial constraint are utilized to find correct correspondences between two feature point sets.Point-wise background regional similarity is introduced to enhance feature points similarity measurement.Meanwhile,based on the adjacent spatial relationship between feature point and its neighbouring points,a new concept of local spatial constraint is further proposed to robustly characterize the geometric consistency between them,which is designed to reduce the ambiguous matches aroused by speckle noise and feature outliers.By combining these two improvements,we generate a new matching cost function and formulate non-rigid image registration as a correspondence optimization problem,which can be solved by the probabilistic relaxation method.Experimental results on both simulated deformed and real SAR images indicate the robustness and effectiveness of the proposed method.5.In order to obtain uniformly distributed feature points and reduce the influence of nonlinear intensity variation to descriptor generation,a novel SAR and visible image registration method is proposed based on nonlinear diffusion and phase congruency structure(PCS)descriptor.Based on nonlinear diffusion scale space,a novel feature detector,namely uniform nonlinear diffusion scale space-Harris(NDSS-Harris),is proposed with Harris detector,proportional coefficient and block strategy.Compared with Gaussian scale space-Harris,NDSS-Harris is able to obtain uniformly distributed feature point in both scale space and image space.In the construction of feature descriptor,the structure map of weighted phase congruency is generated to highlight salient structures in multimodal remote sensing image.To improve the robustness and discriminative ability of feature descriptor,the structure map is divided into different segments according to their phase congruency orders.The PCS descriptor is constructed by concatenating the self-similarity descriptor of those segments.Experimental results on real SAR and visible images indicate that the proposed method is able to efficiently overcome the nonlinear intensity variation,which leads to an improvement on descriptor distinctiveness and registration result.