Heterogeneous Image Matching Based On ROI Features And SAR Scene Matching Navigation

In order to meet the development needs of scene matching assisted navigation technology,this paper focuses on two key technologies:heterogeneous image registration key technology and airborne platform positioning key technology.The paper proposes a registration method for airport targets.Firstly,the Otsu segmentation algorithm is used to extract the region of interest(ROI)according to the target features.Then the Hu invariant moments and K-means are combined to achieve the matching of ROI pairs in heterogeneous images.Finally,the pairs of mismatched points are removed,and the affine transformation model is obtained.The paper proposes a registration method for village targets.Firstly,the village ROIs are extracted by Gray-Level Co-occurrence Matrix(GLCM)algorithm.Next,the Hu invariant moments are used for regional description.Then,the Euclidean distance measurement and the cosine similarity measurement are jointed to achieve the matching of the ROI pairs in the heterogeneous image.Finally,the mismatched pair is removed and the affine transformation model is obtained.The paper proposes a registration method for road targets.Firstly,the adaptive segmentation algorithm is used to extract the preliminary ROIs.Next,the Hough transformation based on the Ratio of Average(ROA)operator is used to extract main directions of roads.Then the linear operator of each obtained direction is constructed to obtain the final road ROIs.Finally,a triangle angle method is proposed to complete the ROI matching between the two images and to obtain registration image.In order to further improve the accuracy of the above-mentioned registration images,an accurate registration algorithm based on phase congruency is proposed.The phase congruency feature is used to extract the feature map of the registration image,and a local template matching method is proposed to achieve accurate registration.Experiments have shown that the accuracy of the final registration image.further satisfies the engineering requirements.Finally,based on the accurate registration image,this paper designs and preliminarily completes the positioning of missile platform,which lays a foundation for the engineering application of key algorithms.