A Fast Iterative Calculation Method for Semi-Dense Stereo Matching

Abstract

The invention relates to the technical field of computer vision, which discloses a fast iterative calculation method for semi-dense stereo matching, which improves the matching speed while meeting the requirement of obtaining higher matching accuracy. This method can be summarized as: extracting the feature points of the left image and the right image respectively, and constructing feature descriptors; then, according to the feature descriptors and epipolar constraints, complete the feature point matching of the left and right images, and the feature points that are successfully matched are called support points ; Next, build a Delaunay triangle in the left image according to the support points, and estimate the prior parallax d of all pixels in the left image i ; and calculate the prior disparity d i The corresponding cost error C i , to obtain the minimum cost error C of all support points in this iteration min ;Continuously iteratively update the support point set and the minimum cost error until the iteration stop condition is met, and finally, calculate its matching point in the right image according to the parallax of each pixel in the left image.

Description

Technical field[0001]The present invention relates to the field of computer visual technology, and more particularly to a rapid iterative calculation method of semi-densified vertex matching.Background technique[0002]Image stereo matches are an important branch of computer vision, photogrammetry, computer graphics, and have a very important value in many applications. Image match can be divided into spars match and dense matching. Sparse matching is typically a feature point having a strong texture on the image, and then calculates the matching cost by the feature description sub-match. Due to the sparseness of the feature point, sparse matching cannot provide sufficient feature points and three-dimensional points in many applications, so it can only achieve more limited three-dimensional world information. The intensive match is to match each pixel point of the image, so the intensive 3D World information can be obtained. Algorithms of intensive match can be divided into global met...

AI Technical Summary

Problems solved by technology

However, due to its inherent computational limitations, dense matching cannot achieve higher processing frame rates (eg, VGA image efficiency of 30fps or higher under equivalent computing conditions)

[0003] Therefore, the above two stereo schemes in the traditional technology have the following defects: sparse matching has high processing efficiency, but cannot provide a sufficient amount of matching point information; dense matching can obtain dense matching point information, but cannot obtain high processing efficiency , unable to adapt to the increasingly high industry requirements

Images