Robust and stable matching method for aerospace images considering weighted Sobel filter feature enhancement
In view of the significant cross-viewpoint differences,scale and size differences,rotation and displacement differences,and nonlinear radiation differences between aerospace images that lead to the stable and efficient problem of matching aerospace images,a robust and stable matching method was designed for aerospace images considering weighted Sobel filter feature enhancement,which can achieve stable matching between satellite images and unmanned aerial vehicle images.Firstly,phase consistency calculation was performed on the aerospace images to obtain the maximum and minimum phase torque map,and feature point detection was performed on the torque map.Then,the feature points were described using a weighted gradient feature descriptor that takes into account weighted Sobel filtering feature enhancement.Finally,using Euclidean distance as a matching measure for identifying Homologous points,whereby multiple sets of aerospace images with varying degrees of perspective differences,rotation differences,scale differences,and nonlinear radiation differences were taken as data sources,and comparative experiments with algorithms were conducted such as the improved SURF detector and localization based on oriented least square matching,histogram of orien-ted self-similarity,position scale orientation-SIFT,and scale invariant feature transform.The experi-mental results show that,in the field of aerospace images matching,the proposed method significantly outperforms the other methods in terms of comprehensive matching performance.So,that method can thus achieve robust matching of aerial images.
万方数据
维普
