Intelligent Inspection System For Engine Parts Based On Machine Vision

Engine is the heart of the automobile that provides the power for the automobile.Its performance directly affects the power,economy and environmental protection of the automobile.The piston ring is the key part of the engine,which plays the role of support,sealing,heat transfer and guidance.Its quality directly affects the efficiency,energy saving and emission reduction and reliability of the engine.At present,the detection of parts mainly relies on the traditional manual or simple visual means,although the simple visual inspection solves the problems of low efficiency,high error rate and large labor intensity in the traditional manual method.There are still many limitations in efficiency and precision.Machine vision,as an online intelligent detection technology,combines image processing,pattern recognition,artificial intelligence,computer vision and optics.It has the advantages of high precision,high efficiency,high degree of automation and programmable,which can solve these problems.Based on machine vision technology and engine piston ring,The size,shape and function of the piston ring has been studied and analyzed in this paper.And combined the algorithm of image processing and detection with related machine vision detection technology is researched to realize intelligent detection.The main contents of this paper are as follows:According to the requirements of inspection task and precision,the hardware system of machine vision intelligent detection is built,including the overall structure design,hardware system design,the working principle of vision detection technology,and the calibration of the system is studied and completed.The principle and denoising effect of adaptive median filter has been studied.Extraction of piston ring parts area is extracted by threshold Segmentation method.The “open” and “closed” operation of the image is carried out by morphology to eliminate the “concave” and “convex” points in the image.The principle,advantages and disadvantages of the image edge detection algorithm are studied,and the edge detection of the image is carried out by combining the Canny operator and the morphological edge detection algorithm.The experimental results show that the algorithm has the advantages of smoothing the edge,suppressing noise,and can detect fine.Complete parts edge,good detection effect.The methods of detecting,discriminating and removing the ghost image of piston ring parts have been studied in this paper.A new method of ZNCC-DWT fast ghost mosaic based on SURF operator is proposed,which includes SURF algorithm feature point detection,enhanced nearest neighbor matching algorithm to judge the correctness of feature point matching.Zero-mean normalization algorithm(ZNCC)is used to judge the ghost image,Data Wavelet Transform(DWT)fusion method is used to remove the ghost image.The Homography matrix(H)is used to realize image stitching.The detection of optical seal density,the removal of image shadow and the effect of image stitching have been studied.Based on the above research,a real-time intelligent software detection system for precision parts based on machine vision is designed by using windows 7 as the operating platform and combing Qt5.7,function library,such as Open Cv and HALCON,and MATLAB,which includes system calibration,image acquisition,image processing,image mosaic,image detection and detection results.Finally,it is applied to the detection of the piston ring of the automobile engine,which realizes the detection of the density of the optical seal of the piston ring and the removal of the shadow between the closed gap and the splicing of the piston ring.The effect of different stitching methods is studied,and the quality of stitched images is evaluated.Using the least square method to fit the piston ring image to obtain the gap width,circle fitting to determine the central coordinates of the piston ring,the diameter of the inner and outer ring,and so on,to verify the feasibility of the system.