Stereoscopic vision-based real low-texture image reconstruction method

Abstract

The invention discloses a stereoscopic vision-based real low-texture image reconstruction method, which is implemented by the following steps: (1) shooting images by two cameras at two proper angles at the same time, respectively, wherein one image is used as a reference image and the other image is used as a registered image; (2) calibrating the inside and outside parameter matrixes of the two cameras respectively; (3) performing epipolar line correction, image transformation and Gauss filtration according to the calibrated data; (4) calculating a self-adaptive polygonal prop window of each point in the two calibrated images, and calculating the matching of pixel points to obtain a parallax space diagram; (5) completing dense matching by executing a tree dynamic programming algorithm pixel by pixel in the whole diagram; (6) extracting error matching points according to a left and right consistency principle, and performing parallax correction to obtain a final parallax diagram; and (7) calculating the three-dimensional coordinates of actual object points according to the calibrated data and a matching relationship to construct the three-dimensional point cloud of an object.

Description

technical field [0001] The invention belongs to the technical field of binocular stereo vision, and relates to the problem of reconstruction of real-shot low-texture images based on stereo vision, in particular to a method of introducing specificity based on pixels in a reference image and reference images and reference images in an adaptive polygon matching window. The dissimilarity of pixel point pairs between registration images is used to calculate the matching degree and the method of obtaining real-shot low-texture image 3D point cloud based on tree dynamic programming algorithm. Background technique [0002] Binocular stereo vision technology is a passive 3D measurement technology. The main tasks to be solved are: camera calibration, stereo matching and 3D information recovery. Passive stereo vision technology can obtain dense 3D point cloud coordinates without projecting auxiliary information such as gratings to the measured object. It has the advantages of friendly ...

AI Technical Summary

Problems solved by technology

[0004] (1) In the calculation of the matching degree, the matching measurement functions such as variance sum, absolute difference sum, zero-mean normalization and adaptive weight measurement are simply used in the rectangular or square support window; if the support window is too small, it is not enough to include Enough grayscale changes to guide the matching and misestimate the disparity value; too large will include points in the same and different disparity ranges so that when the matching degree reaches the extreme value, it does not represent the correct matching position and blurs the boundary information of the region; and the above measurement function In low-texture areas, it may not be enough to distinguish blurred pixels, resulting in many-to-one false matches;

[0005] (2) Simply adopt the traditional dynamic programming global optimal algorithm, the traditional dynamic programming scans each scan line separately, lacks the fusion of horizontal continuity constraints and vertical continuity constraints, and the single-point pixel mismatch will affect the same scan The matching of subsequent pixels on the line produces obvious fringe effects; and when the texture of the gray-scale real-shot image is not sufficient and there is noise, the points to be matched in the low-texture and parallax jump areas cannot be captured enough for correct matching. The correct texture information will easily lead to wrong matching results, resulting in unsmooth reconstruction of point cloud results and unclear boundaries.

[0006] Due to the above shortcomings, the existing real-shot low-texture image reconstruction algorithms cannot obtain satisfactory reconstruction results in practical applications.

Images