Research On Gear Online Inspection System Based On Machine Vision

With the rapid development of computer technology,machine vision technology has become more and more widely used in parts inspection.Compared with traditional inspection methods,this non-contact inspection method has greatly improved the degree of automation,inspection efficiency and inspection.Accuracy,while greatly reducing the labor intensity and cost of testing personnel.At present,there are corresponding visual inspection equipment for gear testing,and the accuracy is very high,but most of them are sampling testing for a single gear,and cannot be tested online during the production process.Most of these equipment are imported from abroad and are expensive.It is difficult to popularize in the industry.In view of the above current situation,this paper through the in-depth study of machine vision technology and image processing technology,taking gears as the research object,analyzes the image preprocessing technology and edge detection technology in detail,and designs a gear vision based on machine vision system.The main research contents of this article:(1)Completed the hardware and software design of the machine vision gear online detection system.According to the specific requirements of gear testing,a suitable light source,camera and other equipment were selected,and a testing platform was set up in a laboratory environment to complete the collection of gear images.The GUI interface design tool of MATLAB is used to design the gear image acquisition interface,gear size measurement interface and gear internal tooth detection interface.(2)Preprocessing the acquired images.The collected original image has a lot of noise,and the image needs to be filtered to obtain an image that is favorable for edge detection.This paper uses Gaussian filtering to filter the image.(3)Research on gear image edge detection algorithm.Through comparative analysis of some classic edge detection algorithms and experimental verification using gear images,the Sobel operator is used to extract the edges of the gear images,but the edge accuracy obtained is low.It is necessary to further subdivide the edges and use robust regression to extract the edges.The first fitting is performed to remove the interference points,and then the second fitting is performed by using the minimum residual absolute weighting method.The experimental results show that the edge can be effectively detected by the two fittings,and the positioning accuracy is improved..(4)System experiment verification and error analysis.The hardware platform and software system are used to complete the gear detection,the gear detection data is obtained,and the cause of the detection error is analyzed.The gear vision online detection system based on machine vision designed in this paper can detect the dimensional data of the gear grooves and the characteristics of the internal teeth in real time.Through experimental verification,the algorithm in this paper can meet the requirements of detection accuracy and speed.