Design And Implementation Of Online Machine Vision System For Engine Fan Cover

With the successive introduction and implementation of German Industry 4.0,China Manufacturing 2025,intelligent manufacturing and smart factory strategies,machine vision is increasingly used for quality inspection and dimensional measurement of industrial products.Automobile engine fan cover is used as an injection molded wedget,and its surface often has various defects such as burrs,irregularities,dents,and skews.In response to the detection of these defects,this paper designs an automatic calibration system and machine vision inspection system based on the basic theory of traditional machine vision image analysis,so that it can detect the defects of the workpiece online.System promptly instructs workers to trim and sort the workpiece.The main work done in this paper is as follows:1)Developed an automated camera calibration equipment.In view of the inaccuracy of the camera to manually capture the calibration image,this paper designs a set of equipment for the combination of pan/tilt and screw,which makes the camera take photo of calibration image precisely.Based on Zhang's calibration method,OpenCV is used to design the calibration program of different types of cameras,which realizes the versatility of cameras such as RGBD,infrared,monocular and binocular,and overcomes the cumbersome process of switching programs between different cameras.2)A machine vision inspection platform was built.A set of machine vision inspection platform for obtaining workpieces,detecting workpieces and processing results from the assembly line was designed.The selection of key sensors and power components for industrial cameras,photoelectric switches,solenoid valves,cylinders and stepper motors has been completed.Completed the three-dimensional design of key modules such as grab structure and camera module.3)Designed and optimized the flow of machine vision algorithms.Based on the traditional machine vision inspection process,the specific engine fan cover of this paper is optimized,and some important algorithm are redesigned,which improves the accuracy of image detection.The filter is optimized.The median filter is mixed into the Gaussian filtering.The hybrid filtering method based on Gaussian filtering is proposed,which overcomes the shortcoming that ordinary Gaussian filtering can not filter the impact noise.The image enhancement algorithm is redesigned to realize the image grayscale expansion algorithm adapting to the overall grayscale of the image,and the details of the workpiece in the digital image are strengthened.Finally,a new circle fitting method based on the nearest neighbor polymerization(KNN)aggregation principle is proposed for the fitting problem of the circle.The fitting effect of the method when the effective pixels are small can exceed the fitting of the Hough transform.4)Designed and developed of machine vision systems.Based on the new machine vision detection algorithm,a complete inspection system was designed.The system is divided into two parts: image acquisition and visual inspection system.The image acquisition system is developed based on the lower computer and hardware sensor.The algorithm part of the visual inspection system is developed based on OpenCV,and the interactive interface of the whole system is developed by using Qt design.5)System deployment,operation test and result analysis.