Research On Measurement Method Of Transmission Case's Mounting Holes Size Based On Binocular Vision

With the progress of society and the booming economy,people's material living standards have significantly improved,and the demand for automobiles is increasing.The challenges and development space for auto parts manufacturers are also increasing.The transmission housing is a vital part of the automobile transmission system.It can ensure the correct installation of the transmission and other connected components.The shape and position parameters of the transmission housing mounting hole are the key to evaluating the quality of the housing.Traditional manual measurement methods are mainly contact measurement,such as vernier calipers and three-coordinate measuring instruments.This method has disadvantages such as large manual intervention,time-consuming,and unfavorable mass production.In response to the above problems,this paper,based on machine vision technology,proposes a method for measuring the shape and position of the mounting hole of the transmission housing based on binocular stereo vision technology.Specific work content includes:(1)Research and complete the comparison of calibration methods,choose a high-precision,low-cost,simple and practical Zhang Zhengyou plane calibration method to complete the camera calibration experiment,and use the reprojection error to evaluate the calibration results,and finally get accurate calibration results.(2)Aiming at the specular phenomenon on the surface of the transmission case,a method combining threshold segmentation and histogram matching is proposed to effectively retain the texture information on the surface of the transmission case and eliminate the specular phenomenon on the surface of the case.After the experimental comparison of the self-made edge test chart,the Canny algorithm is selected to extract the edge of the shell image.Aiming at the shortcomings of the Canny algorithm,such as being susceptible to environmental factors and weak adaptive ability,an improved Canny algorithm is proposed.The adaptive smoothing filter replaces the original Gaussian filter for detection to improve the anti-interference ability of the algorithm;The direction of 45° and 135° is added in the direction to enhance the accuracy of the edge;the maximum inter-class variance method is used to realize the adaptive adjustment of the threshold;the neighborhood search method is used to further connect the edge.Experiments show that the improved Canny algorithm can effectively enhance the denoising performance and improve the detection accuracy.(3)A stereo correction scheme combined with Bouguet algorithm is designed to reduce the feature point search method from two-dimensional to one-dimensional.The row-aligned image is obtained through stereo correction experiment,which speeds up the calculation of stereo matching.On the basis of the traditional SIFT algorithm,the RANSAC method is introduced to remove mismatch points.Experiments show that this improved method can effectively reduce the mismatch rate and reduce the matching time of feature points.(4)Complete the selection and construction of the hardware equipment of the binocular stereo vision measurement system,design and debug the software part of the measurement system.Develop a general plan for the measurement of the mounting hole size of the transmission housing,and complete the relevant experiments according to the guidance of the plan to achieve the measurement of the mounting hole size of the transmission housing.Through the analysis of experimental measurement errors,it is proved that the binocular vision measurement system proposed in this paper can accelerate the detection speed while having more accurate detection results.Finally,the theoretical measurement error was calculated,and the high-precision measurement method was formulated in detail.