Design Of Concrete Thickness Measurement System Based On Binocular Vision

At present,prefabricated concrete structure buildings are the main form of prefabricated buildings in China,which have the advantages of relatively low cost and wide scope of application.They will continue to occupy the dominant position of prefabricated building structures in the future.As the key equipment of the prefabricated component production line,the prefabricated concrete distributing machine's thickness greatly affects the efficiency of the prefabricated component.If the thickness of the concrete cloth is substandard,it will often result in insufficient density or insufficient strength of the precast concrete.Nowadays,most of the concrete thickness is measured by the human eye,and the measurement efficiency is low.However,the innovative laser measurement can only be measured at a single point and is discontinuous.Measurement technology is increasingly not meeting the requirements of industrial automation.In order to solve the problems of blind area in the measurement of the thickness of concrete spreader by human eyes,and that the laser ranging method can only measure the thickness of concrete spreader in a single point intermittently,this paper designs an automatic measurement method based on binocular vision.The hardware part of the concrete distribution thickness measurement system consists of two industrial cameras,camera bracket,computer and high-definition projection lamp.The software consists of five parts: double target location and stereo correction,image preprocessing,feature point detection and description,stereo matching and thickness measurement.After the hardware system is built,the measurement model of binocular vision is established by collecting the chessboard image for double target determination,and the measurement image is corrected by the calibration parameters.The image preprocessing includes the location and clipping of the tested object.In this paper,the corrected image is binarized by local adaption,and the image is processed by denoising,filtering and expanding.Finally,the connected region is detected and the external contour is added with a rectangular frame,and the image is reduced by the rectangular frame.In the description and matching of feature point detection,an adaptive region matching algorithm based on polar constraints and orb features is proposed.The improved orb algorithm is used to force the description of all the points of the image.The described image is stereo matched through the region and the left and right consistency detection is carried out to determine the matching points.Finally,according to the matching results,the distance information of the concrete distribution is calculated by the parallax of the left and right images,and the thickness information of the concrete distribution in the imaging field of view is obtained according to the distance information when the concrete is not distributed.Finally,the concrete thickness measurement experiment is carried out in the laboratory by using the binocular camera and the concrete simulation mold.The experimental results show that the system can be used to measure the thickness of concrete in a wide range and multi-point with high accuracy.The minimum measurement error is 0.436 mm and the maximum measurement error is 1.539 mm.The absolute error of the thickness measurement of the second simulation die is not more than 0.83 mm,which fully meets the measurement requirements of the national standard of not more than 3 mm.