Design And Development Of Machinery Parts Precision Measurement System Based On Machine Vision

Machine vision detection refers to acquire the image information of the target by machine vision method, comparing it with the standard that obtained beforehand, and then ascertaining the quality condition of the target. Machine vision detection overcomes the disadvantages of traditional detection methods, and it is widely used because its high speed, high precision and non-contact. It is one of the hottest research topic of detection technology.In this paper, by taking advantage of photoelectric technology, digital image processing technology and computer technology, taking the mechanical parts of wafer type as the main measurement object, a system of image detection was designed to measuring the size of target. This system consists of lighting system, digital image collection system, image processing system and size measurement system. A suitable image collection system had been designed and it had been calibrated to meet the testing requirements. Going research into the digital image processing technology, includes the early stage of the image process, pixel-level edge detection and sub-pixel edge detection. A software system based on the research had been Developed to measure the mechanical parts size.Firstly, design suitable lighting source and lens construction. Camera calibration technique was researched on to get camera internal parameters, camera external parameters and camera distortion parameters. Correcting the distortion existing in the system to collect the image that meet the requirements of the detection, layed the foundation of accurate measurement of the target size.Secondly, during the image pre-processing color image gray processing, image denoising and binarization method were studied.In order to acquire good results of image pre-processing, an algorithm of image pre-processing is proposed, which was taking use of the weighted average algorithm to change color images into gray images, selecting the median filtering algorithm as the algorithm of image denoising, and binarize the image.Thirdly, the accurate information about the edges of mechanical parts was collected by using edge detection algorithm. During the image edge was detected, various representative edge detection algorithms were realized, and those algorithms are compared with one another by experiments. The principle of sub-pixel edge detection technology was studied, base on the gray moment, an improved gray moment algorithm was presented. The templates of the algorithm were increased to the size by9×9, and the positioning performance of the algorithm was improved. The experimental result shows that the precision and speed of the algorithm were raised effectively.Finally, the accurate size of the target is acquired by recognizing and measuring the edges of the image. Performance of the system is checked by static measurement and dynamic measurement. Results of measurement had been counted to discuss the reason which the error exists in the system. The experiment shows that the measure results of mechanical parts by the system in this paper were accurate and effective, which provides some references to the research on image detection system in the future.