Research On The Geometric Dimension Detection Method Of Mechanical Parts Based On Image Processing

The high precision measurement of mechanical parts geometry and size has been an important part of industrial production and scientific research. Have a direct impact on production quality and efficiency. With the development and progress of modern industry, the traditional means of detection is difficult to fully meet the high requirements of mechanical parts with high accuracy and speed in modern industrial production. Mechanical parts detection method based on image processing is the development trend of modern detection techniques, one that is high-precision, non-contact line detection technique. The advantages is obvious when compared with conventional mechanical testing machine, especially for the small machinery size, large machinery size, dimension detection of complex shaped mechanical structure and profiled mechanical surface.The main significance of this project is to do exploratory research on the measurement of the two-dimensional size of parts with complex geometry in modern machinery manufacturing, and to summarize the common and special problems of image processing technology application in this area. The geometry detection system is established based on image processing and the main context can be devided into the following sections:(1) The commonly used image processing technology in geometric shape and size measurement has been studied and summerized , such as grayscale images, split filtering etc and also the processing techniques, the edge detection techniques etc , along with comparative analysis combined with researched instance, in order to meet the needs of mechanical part dimensional inspection.(2) Based on the study of several edge detection algorithm, the canny algorithm is chosen to get the two-dimensional outline of the part under test fast and effectively, and the subpixel refining method is used to refine the outline information.(3) And the method is proposed to do the edge fitting first and then establish the corresponding relationship with the measured parameters through the fitted part edge. The fitting model of the linear, circular and polynomial curve in two-dimensional profile of the part is improved through research. Fitting of the part edge under test and establishment of the relative relationship between the fitting edge and the parameters tested. And the method is proposed to do the edge fitting first and then establish the corresponding relationship with the measured parameters through the fitted part edge.(4) The test device is built on base of the exsting SVM250Ⅱ-type video measuring machine, two-dimensional images of the part is captured by the SVM250Ⅱ-type video measuring machine CCD camera. Image distortion due to the camera system is corrected using the gridding technic. Parameter measurement is done on the external thread size to verify the algorithm.