Research On Positioning And Recognition System Of Shield Segment Mould Based On Machine Vision

As an important part of urban rail transit,the subway has played an important role in solving the problems of urban traffic congestion,energy conservation and emission reduction.At present,it has become one of the key and hot spots of national public transport infrastructure construction.Among the subway construction methods,shield construction has become the mainstream method of subway construction,in which shield segments play an important role.However,in the production process of shield segments,it still relies heavily on manual operation.In order to improve the automation level of shield segment production,we cooperated with a company on the project of "Research and Development of Automatic Water Receiving and Plastering Equipment for Shield Segment Concrete".This subject is based on the mould offset positioning problem in the project,and the machine vision technology is used to realize the positioning identification and distance measurement of shield segment mould.The experimental results show that the recognition accuracy of the visual positioning method proposed in this thesis is more than 93.9%,and the positioning error is within 2 ～ 3 mm.It meets the positioning accuracy requirements of the project and can be used for industrial field deployment.The specific research contents of this paper are as follows:(1)The machine vision positioning system is designed in detail,and the working principle and composition of the vision system are analyzed.According to the design index requirements,select the hardware of the vision system,and complete the construction of the hardware platform of the vision system.Through the hardware platform,image acquisition,image analysis and data transmission can be completed.(2)The pinhole imaging model of the camera and the causes of image distortion are analyzed.The internal and external parameters and distortion parameters of the camera are solved by Zhang Zhengyou calibration method,and the image distortion is corrected according to the calibration results.A simple method of camera hand eye calibration using camera movement and image processing technology without using plane target is proposed.(3)The image preprocessing algorithm and template matching algorithm based on gradient direction are studied.By comparing the experimental results of different preprocessing algorithms,the most appropriate algorithm is selected as the implementation method of the function module.The extraction principle and calculation method of image gradient feature are analyzed.The working principle and matching process of template matching algorithm based on gradient direction are systematically studied.The shortcomings and existing problems of the algorithm in practical applications are analyzed.(4)The original template matching algorithm based on gradient direction is improved.Aiming at the problem of matching accuracy under nonlinear illumination,the matching priority of strong edges of target objects is improved by giving different weights to feature points.Aiming at the problem of incorrect matching under occlusion,the method of calculating matching similarity in different regions is adopted to reduce the influence of occlusion area on the overall similarity score.Aiming at the problem of calculation speed of the algorithm,the region of interest extraction method and early termination similarity scoring strategy are adopted to reduce the matching search range and calculation.(5)The visual interface of visual system is developed.The interface of each software module is designed based on QT graphics development framework,C/C + + programming language and Open CV image processing interface.Through the software function test,the stability and ease of use of the software system are verified.