Fisheye image matching method based on spin matrix search

Abstract

A fisheye image matching method based on spin matrix search belongs to the field of computer vision research. The fisheye image matching method which adopts spin matrix search between the unit spherical projections of the feature points of two images is not reported in domestic and overseas similar literatures at preset yet. The method first extracts the feature points from the two fisheye images, and projects the feature points onto unit spheres according to the principal points and foci of the images; at least six pairs of corresponding points are randomly chosen from the projected points on the unit spheres to estimate spin matrixes; finally, according to all the spin matrixes, the projected points of the feature points of the first unit spherical image on the second unit spherical image are estimated, the sums of the included angles between the projected points and the corresponding feature points on the second unit spherical image are calculated, and the spin matrix corresponding to the minimum included angle sum is the searched spin matrix. The fisheye image matching method can determine the corresponding points between the fisheye images at the accurate matching rate higher than 80 percent by way of spin matrix search between the fisheye images.

Description

technical field [0001] The invention relates to a method in the field of computer vision, in particular to a fisheye image matching method based on rotation matrix search. Background technique [0002] The 3D reconstruction technology that restores the visible surface geometry of 3D objects from 2D images has always been an important research content of computer vision. In recent years, with the continuous development of applications in digital museums, digital archaeology, virtual reality and other fields, the visual effects of object surface structures have become more and more concerned about 3D reconstruction technology, and the matching method of image corresponding points is to apply 3D reconstruction A key issue of technology. Currently commonly used matching algorithms such as feature point-based, Harris corner-based, and correlation-based matching algorithms are all aimed at perspective images, but there are few reports on fisheye image matching methods. Due to it...

AI Technical Summary

Problems solved by technology

The above fisheye image matching methods are all carried out on the fisheye image plane, and do not make good use of the projection point information of the feature points on the unit sphere on the fisheye image

Images