Research And Application Of Method Of Etracting Acuracy Object Edge Parameter In Size Detection Based On Computer Vision

Computer vision can replace human vision in inspection, target tracking, robot guidance and other work, especially those who need repeated access from the images quickly and accurate information of the occasion. Computer vision in automatic production line can replace the artificial in the rapid, accurate and monotonous work product inspection. With the rapid development of computer technology, computer vision in medical diagnostics, industrial automation monitoring and control, intelligent robot navigation and target tracking and other fields are widely used, especially in the apparent quality of the manufacturing test, measurement, quantitative rapid detection of other aspects of development, has made tremendous economic and social benefits.Using computer vision system to detect the size of engine valves, valve can be achieved on the rapid, non-contact, high accuracy detection and pipeline operations, but also provide users with convenient, rich control interface, and further enhance the level of automation.The main subject of computer vision in the existing system of early detection object component size flat-field compensation, calibration and measurement algorithms such as lens distortion correction, based on further study of engine valve parts collected images of the horizontal, vertical, diagonal and circular, etc. edge detection method, using the software program to extract the exact parameters of the edge, accurate size measurement and practical applications. The valve size of the practical application of computer vision testing, the proposed method proved accurate and stable, high-precision visual inspection to meet the requirements.Contents of the study include:1. Using Prewitt edge operator to complete the initial position, on this basis, through the change of edge discrete values for gray squares curve fitting, and the fitting curve extremum, get the edge accurately.2. Proposed size of the known multi-shaft as the reference model to establish error distortion correction, automatic calibration using standard parts, making the system more stable, more reliable measurement data. 3. The actual detection of a variety of valves, achieved good results, reachingμlevel testing requirements.