Precision three-dimensional vision measurement method based on binocular camera and structured light projection

Abstract

The invention relates to the field of three-dimensional scanning and measuring technology and calculating stereo vision, in particular to a precise three-dimensional vision measuring method based on binocular camera and structured light projection; the method comprises: setting up a binocular camera and a digital optical projection device, and calibrating a binocular camera; generates a structuredlight texture according to a multiphase cosine texture and gray code algorithm or a multi-frequency multiphase cosine texture algorithm and burns it into a digital optical projection device; projectsa structured light texture using a digital optical projection device, and simultaneously acquires a texture image by a binocular camera; Decodes the texture image, and restore the true phase value byusing the median filtering based phase correction algorithm; according to the polar line based phase binary search algorithm, locate the pixel points in the a stereo cameras having the closest absolute phase values; use the binocular to demarcate the result, and generate a point cloud by the inverse triangulation. The invention combines the structured light projection with the binocular camera, and uses a plurality of algorithms to realize precise measurement of the image with high precision.

Description

technical field [0001] The invention relates to the fields of three-dimensional scanning and measurement technology and computational stereo vision, in particular to a precise three-dimensional vision measurement method based on a binocular camera and structured light projection. Background technique [0002] 3D scanning and measurement technology is widely used in manufacturing, electronic circuit industry, art and entertainment industry, archaeological industry, medical industry and so on. In these industries, it usually involves high-precision three-dimensional scanning of a certain type of object, and analysis and processing of the obtained point cloud model, common defect detection, 3D modeling, etc. This type of 3D scanning is characterized by short scanning times, high point cloud accuracy, and excellent preservation of details. [0003] The traditional multi-camera multi-view geometric 3D modeling method generally finds the corresponding relationship between the ima...

AI Technical Summary

Problems solved by technology

[0004] 1) Due to the noise generated by the image, the surface of the point cloud generated by the calculation is too rough;

[0005] 2) The absolute phase value at the relative phase discontinuity is prone to sudden jumps. Generally, neighbor interpolation or mean value filtering is used, but details will be lost and artificial feature defects will be introduced;

[0006] 3) Using horizontal and vertical bidirectional textures takes a long time to acquire images, and using two-dimensional search methods to search for matching features violently, the computational complexity is too high;

[0007] In addition, a single-projection single-camera configuration is also commonly used, in which the projection is calibrated as a reverse camera device, but the calibration process is too complicated, and the projection distortion parameters are difficult to determine. At the same time, the binocular camera or the projection camera is not fully utilized. Geometric constraint relationship, insufficient measurement accuracy.

Images