Research And Development Of Automobile Brake Pad Appearance Quality Inspection System Based On Machine Vision

As an important component of automobile brake system,the quality of automobile brake pad directly affects the driving safety.Its appearance quality is one of the important aspects,and its quality inspection is also very important.The appearance quality inspection system of automobile brake pad based on machine vision is to realize character recognition and bayonet size measurement of brake pad by means of machine vision,which is an important method to realize the automatic quality inspection of automobile brake pad.This paper studied the quality inspection method of the lighting system brake pad appearance quality inspection,character recognition algorithm,dual camera bayonet size measurement system and its algorithm.The main research contents and conclusions are as follows:1)The quality inspection method of the lighting system brake pad appearance quality inspection.Based on the analysis of the correlation between the seven characteristic indexes of gray scale and the mean value and standard deviation of Illuminance,it is determined that the mean value of illuminance is represented by the degree of gray distribution V,and the standard deviation is represented by the gray distribution G The lighting system design quality inspection is realized by using the inspection model based on he degree of gray distribution degree V and the gray distribution G fusion.The test results show that the double-index fusion test model can realize the quantitative test of lighting system design quality effectively,and the correct rate is 90.99%.2)Character checking algorithm.The image is pre-processed by OTSU algorithm,gauss filtering and dilation and erosion,extracted location hole region by BLOB analysis,circle fitting by least squares,and image rotation correction by affine transformation based on the feature of the brake pad,the character region is located.After comparing the results of contour edge extraction with the whole character extraction,the adaptive threshold segmentation method is used to extract the whole character.In order to avoid repeated checking through image filling,we can judge whether the character imprint is correct or not according to the number of remaining black pixels in the image and character sorting.The experimental results show that in this study,the speed of character checking is 1.900 s/image,and the checking accuracy is 100%.3)Measurement of the size of brake pad bayonet based on dual-camera.According to the principle of the distance transformation between the image space and the actual space,the principle of the calibration and the size measurement is analyzed,and a dual-camera size measurement hardware system is designed to collect the image of the standard device and the brake pad,and the edge is extracted by Sobel operator The calibration and size measurement are carried out according to the corresponding principle by least squares fitting.The measurement errors cause by11 measurement variables are analyzed,the error transfer coefficients are calculated,and the error transfer model is established.The synthetic error is 0.0131 mm.The test results show that the error of 100 sets of images are all within ±0.02 mm,which can meet the precision requirement of bayonet size measurement.The above research results provide a new inspection method for the quality inspection of machine vision inspection lighting system,and lay a certain method and technical foundation for the character check and bayonet size measurement of automobile brake pad through the method of machine vision,which contribute to the automation of the appearance quality inspection of automobile brake pad.