Local stereo matching method based on multi-feature integration

Abstract

The invention discloses a local stereo matching method based on multi-feature integration. Firstly the Harris feature points of an input image are extracted, the matching feature points of high accuracy are obtained through feature matching, and the parallax of the matching feature points is calculated and the parallax range is estimated. As for the non-feature points, the joint matching cost is constructed through combination of the LAB color space information of the image, the invariant moment of the image and the gradient information of the image, and region matching is performed by using the calculated parallax search range so as to obtain the overall dense parallax map. The method is low in complexity, high in matching accuracy and suitable for the occasion of high real-time requirement.

Description

technical field [0001] The present invention relates to digital image processing technology, more specifically to a local stereo matching method based on multi-feature union. Background technique [0002] Stereo Matching has always been a technical difficulty in binocular stereo vision. The purpose of binocular stereo vision is to simulate human eyes to obtain three-dimensional scene information, obtain two-dimensional images of the scene from two angles, perform stereo matching and reconstruct a three-dimensional model. The goal of stereo matching is to find the matching corresponding points from different viewpoint images to obtain the disparity map, and then obtain the depth of field of the object through the triangulation relationship. It is widely used in medical imaging, 3D reconstruction, robot vision navigation and target tracking, etc. field. [0003] Most current stereo matching algorithms include four steps: (1) matching cost calculation, (2) cost aggregation (c...

AI Technical Summary

Problems solved by technology

The size of the support window affects the matching accuracy. If the window is too large, the aggregation process of the matching cost will contain neighborhood information that does not have guiding significance, resulting in mismatching in discontinuous areas of disparity. If the window is too small, the matching cost will be differentiated. Too low will increase mis-match in weakly textured areas

[0007] At present, the defects of binocular stereo matching technology are mainly reflected in the lack of matching accuracy and real-time performance, and many algorithms are difficult to take into account the dual requirements of precision and real-time performance

Images