Remote-sensing image matching method with combination of characteristic segmentation with topographic inequality constraint

Abstract

The invention relates to a remote-sensing image matching method with combination of characteristic segmentation with topographic inequality constraint. The method comprises the following steps: partitioning a base image so as to obtain base image blocks; extracting the characteristic points of each base image block; calculating local contrasts of all the characteristic points; selecting a plurality of characteristic points with the large local contrasts as seed points; performing projection on the seed points so as to obtain a corresponding search region of each base image block on an image to be matched; calculating a scale difference factor and a rotation matrix of the base image and the image to be matched and balancing the scale and the rotation difference; introducing in data based on global SRTM (Shuttle Radar Topography Mission); performing hierarchical matching on each base image block based on the seed points and eliminating miss matching points; performing image segmentation and combination; and extracting edge characteristics and segmented regions, thereby obtaining dense matching point clouds. The method has the advantages that the remote-sensing image matching in different resolutions and different perspectives can be realized in a full-automation manner; and the reliable and dense homonymy points can also be obtained from the difficult-to-match regions which are poor and similar in textures, have discontinuous topographic surfaces and the like.

Description

technical field [0001] The invention belongs to the field of surveying and mapping science and technology, and relates to a remote sensing image matching method combining map segmentation and terrain relief constraints, which is mainly used in the fields of multi-source remote sensing image matching, digital surface model generation and the like. Background technique [0002] Image matching, as the core technology in the field of digital photogrammetry and computer vision, has been at the forefront of domestic and foreign scholars' research in the past ten years. Image matching is the process of identifying points with the same name between two or more images through a certain matching algorithm. [0003] The current image matching methods can be roughly divided into three categories: matching based on region grayscale, matching based on features and matching based on semantics. Image matching based on regional grayscale refers to comparing the similarity of the grayscale d...

AI Technical Summary

Problems solved by technology

[0009] In general, there are still some insurmountable problems in the existing matching methods, such as: ① The image must have rich grayscale information (high contrast, high contrast), poor texture information, similar textures, and discontinuous areas on the surface Difficult to match; ②The images must have a similar scale; ③The rotation angle between the images should not be too large; ④The geometric and radiation distortion between the images should not be too large; ⑤The shadow or occlusion range of the images should not be too much, etc.

Images