The Research On The Detecting System Of External Thread Surface Defect Based On The Machine Vision

The application of screw thread in industrial production, mechanical movement, automobile industry,aerospace and so on is very broad. Screw threads are used in different function by the different dimension and materials. Connection, transmission and tightening equipment are the primary function of screw thread. Because of its use is so wide and its role is so important,so to evaluate the quality of screw thread is very significance. In order to keep up with the development of industrial production,we should find a fast and accurate method for testing on the quality of screw threads. The method must find the screw threads with defects quickly,and abandon them to avoid unpredictable serious consequences. The determinants of the quality of screw threads are the dimension of screw threads, surface defect, mechanical properties, chemical composition, surface coating and so on. So far, there is no mature detection method for screw thread surface defect exist at home and abroad. So to develop a detection system for screw thread surface defect is very necessary.After in-depth studied on machine vision technology,a detection system for screw thread surface defect was designed. The system mainly contains two parts,one is Image acquisition,the other is image processing. The following are the main contents of this paper.① Consulted the Related literature in and abroad, and compared the advantages and disadvantages of some NDT(Nondestructive Testing) methods, the detection method based on machine vision was adopted.② In-depth studied on machine vision technology, the image acquisition scheme with area array camera was selected finally,and the selection basis was provided. Based on the collection scheme which was determined,the optical lens, light source, and other devices were selected. The motion control platform for detection system was designed, and the platform of image acquisition had been set up by the selected devices.③ According to the texture features of the screw thread, the processing algorithm was designed to detect the defects. When got the surface image of screw thread, firstly analyzed the noise may existing in the image and used median filter method to filter out the noise. Then used discrete cosine transform to enhanced the image,during the process of enhancing the image, a suitable power filter with the best threshold value which was get by the method of finding the maximum of curvature was designed to filter out the background information. The filter maximized the contrast between the background information and the defects, thus implemented image enhancement. The defects can be segmented from the enhanced image by the method of SPC(Statistical process control) binary segmentation. Finally based on the judgment method which was given before to judge whether the area was defect or not.④ The PC interface of detection system was designed, it make the system more convenient.Experiments show that the detection system designed in this paper can effectively detect the screw thread surface defects. The defect located on the crest or root of the screw thread can both be detected quickly and efficiently. Because the system has the characteristics of good real-time performance, high detection speed, and non-contact, so the detection system is of great significance.