基于隐含特征和SIFT方法的SAR图像多尺度配准OA

In view of the uneven distribution of feature points and insufficient matching in the process of SAR(synthetic aperture radar)image registration,an SAR image multi-scale registration method based on implicit features and SIFT(scale-invariant feature transform)method is proposed.After polarization decomposition of SAR images,the Wishart distribution is used to describe the coherence matrix gradient of SAR images,and then resolution units 1 to 2 are used to describe the Wishart gradient of SAR images,obtaining a single-stage SAR image ratio gradient.This ratio gradient is the implicit feature of SAR images.The SIFT method is used to establish an SAR multi-scale space,and a downsampled image of SAR images is generated in this multi-scale space.The single-stage SAR image ratio gradient is calculated on the basis of the downsampled image.According to the implicit features of SAR images,the feature extremum points and the main direction of the feature points are determined,and uniform SAR image multi-scale registration feature description vectors are generated.And then,the distances between the feature description vectors are described with Euclidean distance,so as to realize multi-scale registration of SAR images.The experimental results show that the proposed method has strong ability to extract implicit features from SAR images,and can achieve high-quality multi-scale registration in the presence of scaling and rotation of SAR images,with good applicability.