Research On Geometrical Parameters Of The Valve Measurement Based On Machine Vision

Nowadays, with the prosperity and development of automobile manufacturing industry, in order to pursuit the products with higher quality, auto parts processing and manufacturing also ushered in the golden age of its own development. In face of the increasingly requirement of the accuracy of the auto parts, traditional measure technology not only has low detection efficiency, but also cost a lot of labor, high cost of quality and management, it already can not adapt to the development of the trend of manufacturing because of the precision has been able to meet the demand of current production. The rising computer vision gradually occupies the main position in the field of production line of automatically detection, relying on the rapidly development of photoelectric technology and image processing technology with the advantages that it has fast detection speed, contactless measurements and measurement accuracy.The issue makes the sizes measurement of automobile engine valve as the research object and designs the measurement methods of the sizes of the valve on the base of machine vision, then build the testing platform. The main research contents are as follows:Firstly, design the size detection project according to the actual precision of the valve sizes detection system, through analysis and contrast the structure and parameters of the hardware devices, choosing these visual measurement hardwares include the right camera, telecentric lens and light source with a parallel light. In order to get the image of valve with high contrast and sharp edges, emphatically study the lighting solution then design the lighting system with better efficiency through comparative trial, to complete the whole system hardware platform structure.Then, before measuring the sizes of valve, it is necessary to calibrate the camera firstly. The camera used in machine vision need to take into account the convert of the existing realities and it’s images and the problem of lens distortion. Firstly, discuss the camera calibration in detail. Secondly, use HALCON to calibrate the camera in order to get the camera internal and external parameters. Using these parameters can we get the matrix from the existing realities of the world coordinate system to the image point in the image coordinate system, and then get the actual distance of each pixel. Judging the measuring system whether reach the precision requirement through the analysis of the precision.Finally, in view of the problem that traditional method used in pixel edge detection can not orient the real edge exactly, through deeply analysis the sub-pixel edge detection methods include moments method, interpolation method and fitting method. After contrasting the above methods in terms of measuring accuracy, speed, complexity and suitable applications, choose the sub-pixel edge detection method based on Zernike moment to measure the sizes of the valve, and then ameliorate the algorithm for the purpose to improve the precision of the edge position. Finally use the improved algorithm in the measurement of radial dimension, axial size and oblique size and finish the measurement of the sizes of valve. Thus, it is confirmed that the measurement results meet the engineering accuracy requirement by the range of the measurement results.