Detection Of Bridge Apparent Defects Based On Binocular Vision Principle

Apparent defects detection of bridge structures is an important content of the evaluation of bridge health status.The traditional detection method is direct detection method,which uses measuring tools such as tape measure to measure directly.The detection accuracy of this method is not high,which is greatly affected by the environment,and the detection results are also affected by subjective factors.Based on the current situation of domestic inspection,this paper takes improving the speed and accuracy of defect inspection as the starting point of research,and proposes a binocular vision intelligent fast inspection method that uses computer to intelligently identify defects.Through actual bridge defect inspection and laboratory simulated underwater environment inspection,the practicability and applicability of binocular vision technology in bridge apparent defect inspection are verified.The research contents of this paper are as follows:(1)Theoretically analyze the camera imaging model and calibrate the camera parameters based on Zhang Zhengyou calibration method.The image preprocessing method is deeply studied,and Gamma transform and median filter are used to remove brightness change and noise of the original image respectively.(2)Research on the principle of three-dimensional coordinate calculation of spatial points,demonstrate the experimental cases of coordinate calculation of spatial points,and obtain the accurate results of spatial coordinate calculation of spatial points,and deduce the basic algorithm of apparent defect evaluation through mathematical geometry knowledge.The SIFT algorithm is used to extract and match the feature points of the original image,and the spatial coordinates of the feature points are calculated.The matching test results of the feature points are one-to-one correspondence,and the matching accuracy is high,which verifies the feasibility of the SIFT algorithm.(3)In order to verify the applicability of binocular vision technology in the detection of apparent defects of bridge structures,different defects are tested in different environments(air and water).The maximum error is 6mm and the maximum relative error is 25% when different defects are detected in air.In turbid underwater environment,the maximum error is8 mm and the maximum relative error is 40%.In clean water environment,the maximum error is 10 mm and the maximum relative error is 50%.The experimental results show that binocular vision technology can accurately measure the geometric features of the defects on the premise of obtaining high-quality images by the camera.Although the relative error is large,the error is relatively small,thus verifying the practicability and applicability of the technology in the detection of apparent defects of bridges.Based on binocular vision principle,this paper realizes the accurate recognition of geometric features of apparent defects.The main contents include: camera calibration,imagepreprocessing,feature point extraction and matching,three-dimensional coordinate calculation of spatial points and other processes.The test results have reached the expected values.The research results in this paper can basically realize the automation and high precision detection of bridge apparent defects,and have certain development space and application prospects.