Design On The Detection And Identification System Of Mechanical Parts Based On Machine Vision

Machine vision detection and identification technology has advantages as non-contact, good real-time, faster rate and anti-interference ability, etc, which can achieve the goal of zero defect production of mechanical parts, adapt to the requirements of progress and development of modern manufacturing and shows broad application prospects in real production. The main work of this article is as follows:(1) After briefly introducing the history, status and development trend of domestic and international of machine vision, we propose a mechanical parts detection and identification system overall program and describe full range of image acquisition system from hardware and software aspects, including the choice of camera, LED light source selection and image acquisition card selection and analysis, and set up reasonable experimental platform, the initial realization of the system design goals.(2) Using a variety of image processing technology to process the image obtained, make the image more in line with the requirements of the human eye, including image enhancement and sharpening, image denoising and image segmentation, and analysis the pros and cons of various image processing techniques and algorithms in this experiment to determine which method is more suitable for the designed system. (3) We use neural network algorithm for detection and identification of mechanical parts, distinguish the various defects of the mechanical parts and describe the recognition accuracy of various defects, analysis the error of the system and design a friendly software interface finally.