Research On Crack Information Ranging And Location Technology Based On Binocular Vision

As one of the research hotspots in the field of computer vision,binocular vision technology is widely used in crack detection,medical imaging,underwater navigation and other fields.Many scholars have studied binocular vision technology in depth.Although the technology has achieved non-contact measurement,real-time measurement,and high-precision measurement in crack detection,it has been dealing with uneven illumination,low-texture structure,depth discontinuity,and non-contiguous measurement.The measurement accuracy is low when the area is blocked.This paper builds a crack information ranging and positioning system based on binocular vision.The system combines computer vision technology with crack detection in aqueducts to realize the distance measurement,three-dimensional positioning and scale measurement of cracks.Finally,the experimental data is compared,analyzed and summarized.The specific research work of this paper is as follows.First of all,this paper adopts an improved algorithm of median filtering in the process of image preprocessing.The algorithm automatically adjusts the window size according to the gray value of the image pixel during the filtering process,obtains the gray value of the center pixel of the filter window and the neighboring pixels,and uses the variance of the two gray values as the basis for judging noise points.This effectively avoids the misjudgment of noise points in median filtering,prevents signal points from being eliminated,and preserves the detailed information of the image and strengthens the denoising effect of image edge information.Secondly,use the MATLAB toolbox to perform stereo calibration on the binocular camera.The process includes reading images,corner detection and solving the parameters of external and internal the binocular camera.In order to ensure the accuracy of the data source,12 sets of images have been collected.By analyzing the experimental results,the accuracy and efficiency of the stereo calibration are proved.Third,in the stereo matching stage,the neighborhood window is used to perform the complementary color mean decolorization Census transformation,which solves the problem of excessive dependence on the center pixel,and uses the left and right consistency check and disparity refinement to obtain the final disparity map.Through simulation and comparison with other algorithms,the matching accuracy of images in low texture,depth discontinuity and non-occlusion areas is effectively improved.Finally,based on the improved algorithm,a binocular vision crack information ranging and positioning system was built.The system compares the final experimental data with real data through multiple aqueduct crack distance measurement experiments,crack positioning experiments and aqueduct crack scale measurement experiments,which proves that the binocular vision technology has a certain reference and practicality in the field of crack detection research.