Research Of Technology On Surface Defects Detection For Steel Plate Based On Machine Vision

As an indispensable and important kind of raw and processed materials, the quality of steel plate decides the industrial development directly, so it is very significative to control its surface quality. Traditional detection methods of surface defects of steel plate not only comsume large amounts of labor, but also detect the surface defects at a low efficiency and poor quality. Therefore, It is urge to develop an automatic surface defects detection system for steel plate in domestic steel enterprises. Machine vision detecting technology based on the computer science uses machine to carry out the automatical detection instead of manual labor. Using this technology to detect surface defect for the steel plates not only achieves automatic detection and reduces the labor cost, but also raises the efficiency and improve the quality. Therefore, the research on surface defects detection for steel plate based on machine vision technology has become an hotspot in many domestic universities and steel enterprises.This dissertation which takes the scratch defect of steel plate surface as detection object, does researches on some key technologies of disfigurement detecting of steel plate surface based on machine vision. Through this dissertation, the following achievements have been made:1. Rest on the analysis of selecting principles of camera and lens in machine vision industrial detection applications, Cameras and lenses for detection are determined by calculation in accordance with accuracy requirement about the static detection of scratch defects on the surface of steel plate in laboratory. The result after image processing shows that the selected camera and lens satisfy the accuracy requirement.2. Making a study of the illumination technology of the scratch defects detection of steel plate. Integrating both light source and illumination mode theory in machine vision light source illuminating system. Taking LED as light source, Multi-group illuminating experiments have been carried out in different conditions which are high angle, low angle, bright-field and dark field. And sum up the effect of scratch defect detection of steel plate in different illumination modes by means of experimental data analysis, which provides gist for lighting module of the scratch defects detection system for steel plate.3. Rest on the analysis of zhang's algorithm about camera calibration, proposing an improved algorithm to calibrate the camera CanonEos450D. Compared to the traditional zhang's algorithm. The improved algorithm imports tangential distortion, considers the influences of the radial distortion and the tangential distortion, consummates the calibration model, and increases the calibration precision.4. The dissertation studies the image detection methods of the surface scratch defect of steel plate, and uses the methods of image gray processing, image filtering, image enhancement, edge detection and morphological image processing to extract the scratch defect and to measures its length.