Size Measurement And Surface Defects Detection Of Magnetic Shoes And Systematic Development

In recent years, with the constant improvement of the economic technologies of China, more enterprises aspire to gain quality products with high precision and reliability while the common detection methods cannot meet the requirements to detect product fast. With the gradual development of imaging and computer technologies, detection system based on machine vision is realized and is widely used in many lines. In the line of industrial production, many enterprises control the quality of products taking the superficial defect detection system based on machine vision as a tool. Considering the above background, this thesis makes a study on the measurement of magnetic shoe size and the visual detection method on superficial defect and proposes the algorithm of visual detection on the superficial defect and overall size of a magnetic shoe and applies the algorithm into the defect detecting system.Firstly, the defect detection system is designed and made according to the general requirements of the program and actual production environment, which includes mechanical system, visual system, control system and application software system. The mechanical system is mainly composed of transmission actuator, feeding actuator, eliminating actuator, turnover actuator and camera collection actuator among which the elimination actuator mainly eliminates the unqualified magnetic shoes being tested on the production line, the turnover actuator turns over the magnetic shoe and detects defects on the positive and opposite sides of the magnetic shoe; visual system is mainly composed of touching sensor, camera, lightning source and shots, bowl like lighting source of Shanghai Yingcheng Co., Ltd. is adopted based on the arc superficial form of the magnetic shoe so that the light can spread on the surface of the magnetic shoe in a balanced manner. Meanwhile, the camera Genie with the resolution of 1280′960 from Teledyne DALSA of Canada is adopted based on the field of view and the precision of 50mm.Secondly, this thesis studies the pre-processing method of the image of magnetic shoe being collected, this thesis compares the two sorts of filtering algorithms in a comprehensive way and adopts high-pass filtering template to make filtering treatment to the image of the magnetic shoe referring to the result of experiment in a comprehensive way. Meanwhile, as to the marginal abstraction, 3 algorithms are compared before Canny operator is selected. Besides, the OTSU algorithm is adopted to make threshold segmentation of the image and eliminate the possible noises exist after the segmentation through digital morphology.As only one circle can be fitted for the measurement of the external size of the magnetic shoe, this thesis designs the concentric circles fitting algorithm based on ellipse and adopts this algorithm to fit the upper and lower arc of the side of the magnetic shoe in order to calculate the thickness of this magnetic shoe. Additionally, the minimum enclosing rectangle algorithm based on the center of gravity is adopted to calculate the length, width and height of the magnetic shoe. Meanwhile, on the detection of the defects on the magnetic shoe, this thesis designs the classifier with rectangularity, circularity, center of gravity and the length-width ratio of the enclosing rectangle as the eigenvectors to classify the defects of the magnetic shoe. Moreover, the thesis takes whether the size measurement and defects detection meet the standards at the same time to judge whether the magnetic shoe is qualified or not. According to the test, the defects detection and size measurement which is realized within 200 ms meet the requirements.Finally, the software realization of the detection system is interpreted, the image acquirement and processing of this system is respectively based on the development of Sapera LT software base and Opencv open source image processing function base. Meanwhile, the system is based on the development of the integrated environment of Visual Studio 2010 and realized by the MVC framework of MFC framework. The system views dialog class and view class as Views and the corresponding image processing algorithm is regarded as Model, thread control related contents belong to Control class which are beneficial to the follow-up development and reuse of codes. The system adopts independent thread to accomplish the major tasks of image acquirement, processing and display. According to the actual running result, the software runs steadily for a long time.