Research On Gear Defects Detection Based On Machine Vision Technology

Gear is a commonly used and very important transmission parts in various types of mechanical equipment, so the gear’s quality is critical for entire mechanical device, once the defective gear is being used, the heavy load and stress may lead to damage of gears, and lead the fault and scrap to the mechanical equipment, even threaten the oper ator’s safety. So it is particularly important to detect the machining quality of the gear i n the whole process of manufacturing. The traditional gear quality inspection mainly ad opts the method of artificial detection, which is time-consuming and hard to ensure the detection quality. Machine vision inspection is a method using the machine to replace t he human to do measurement and judgment, with the detection speed, detection level st ability and other advantages, so the machine vision detection has become a more popul ar detection method in recent years.This topic comes from the fund project:Guangdong provincial science and Technology Department of industry and new high-tech research and guidance project "flexible manufacturing technology research and development and application of flexible manufacturing technology" 2013B010102022. According to the actual requirements of the enterprise, using the theory of machine vision and image processing, the key technology of the gear defect machine vision inspection system is studied. Using OpenCV to prepare the gear defect detection software, used to replace the traditional manual inspection methods, and promote the development of machine vision detection technology.Contents of this paper are as follow:The significance of the research is discussed, and the current defect detection methods are compared. Then, it briefly introduces the research status of the detection technology, explains the source of the topic and the purpose of the research, and points out the key technology of gear testing. According to the technology and requirements of machine vision inspection, the hardware selection and the selection of software development environment are introduced, and the platform of defect detection system is established.Starting from the type of gear defect detection, the image pretreatment of the gear image which needs to be detected is carried out. Using Gauss filter, median filter, Wiener filter and other filtering on the gear image smoothing and denoising experiments and analysis of the results, it is concluded that the Wiener filter to the gear defect image denoising effect is the best. In edge detection of gear image using Canny edge operator, krisch edge operator, LOG operator operator segmentation experiments carried out, comparing different operator segmentation effect, decided to choose LOG operator to the image of a gear edge segmentation.Finally, focus on the defect detection and recognition algorithms, the gear due to machining marks caused by the interference problem, based on the case of the high rate of false positives and gray level registration in gear defect detection, through to gear defect of log edge detected defect of gear image edge with no defect of gear image edge is very different, therefore, put forward using gear edge difference comparison, to achieve the gear defect detection, selected using concept based on OpenCV contour tree matching of contour matching method, the gear defect detection problem can be solved, and OpenCV programming is realized. Of gear, pits, chipping, missing teeth, deformation, sticky powder defect of recognition experiments, verified based on OpenCV contour tree matching method can effectively identify the gear fault and solve the practical problems encountered in the defect detection has certain guiding meaning of meaning and inspiration, enlarges application range of a contour tree matching method.