Matching method of homonym points in heterogeneous remote sensing images considering high-order structural features

Abstract

The invention belongs to the field of surveying and mapping science and technology, and provides a method for matching points of the same name in heterogeneous remote sensing images in consideration of high-order structural features, including data preparation before heterogeneous image matching; dividing and extracting feature point clusters; feature points Determination of candidate matching points; matching of points with the same name in consideration of high-order structural features; for each layer of pyramid image matching results, using the RFM model block adjustment method of fusion weight selection iterative method; refining the matching results layer by layer until the original image layer is completed , and finally realize automatic and reliable matching of homonymous points in heterogeneous remote sensing images. The present invention comprehensively uses the rational function model and the hypergraph image matching model, and introduces geometric constraint features and high-order structural features into layer-by-layer pyramid image matching, which not only ensures the sparsity of the hypergraph model, but also makes the hyperedge sampling The information loss is minimized; the invention improves the reliability and success rate of heterogeneous remote sensing image matching, and effectively reduces the workload of manual measurement of points with the same name.

Description

technical field [0001] The invention belongs to the field of surveying and mapping science and technology, and relates to a method for matching homonymous points of heterogeneous remote sensing images, in particular to a method for matching homonymous points of heterogeneous remote sensing images in consideration of high-order structural features. Background technique [0002] With the rapid development of earth observation technology, various types of sensors have sprung up. They record the changes of the earth's surface with higher time resolution and spatial resolution, and also make the images covering the same area more and more diverse. . The joint target positioning of heterogeneous remote sensing images has become one of the important research directions in the field of photogrammetry and remote sensing. However, whether it is single-point 3D positioning or block adjustment, it is necessary to quickly and accurately find the same name point of the image and measure ...

AI Technical Summary

Problems solved by technology

However, the focus of its research is to integrate the beam block adjustment method based on the RFM model into the same-name point matching process, use the geometric constraints provided by the RFM model, improve the prediction accuracy of the initial point position of the feature points, and provide epipolar line geometric constraints to narrow the matching The search space and the block adjustment based on the RFM model eliminate the wrong matching points, but it is still a single-point matching in essence, and the high-order structural features between the matching points are not introduced in the matching process

Images