Study On Defects Detection Of Surface In Steel Strip

Surface quality is one of the most important quality parameters for steel strip, with the development of technology, which get more and more concerns from the follow-up industry. The iron and steel corporation pay more and more attention on how to detect and control the defects of surface in steel strip. With the improvement of the computer-vision and computer capacity, automatic surface detection in steel strip based on image has attracted more attention.An experimental device using CCD matrix camera to detect defects of surface in steel strip was designed based on the deep investigation and study of existed surface detection technologies in this paper, and a new system for defects detection was invented. The characteristic of new system are as follow: LED light source has been adopted to illuminate, keep the illumination in the range of 70-100 Lux, which assure the high quality of surface images in steel strip;The appropriate CCD matrix camera which operation area is 0.5mm×0.5mm has been used to capture the small defects at the speed of 5m/s;Image processing, such as filter smoothing the noise, image edge detection, and image mathematical morphology processing techniques applied in surface detection are studied and developed, according to the emphasis on precision requirement, which were more suit for the features extraction;Features extracted by the system are discussed, and the back-propagation network has been chosen to as the method of image classification, they was efficient of processing defects image of surface in steel strip;And the simulation experiment was carried out with the experimental device. The result shows that main types of surface defects, which were investigated of this paper, can be detected, and the average detection rate is 91%.This kind of detection method is novel with simple equipment, which can meet precision requirement of most product line of steel strip company and have a greatpotential to popularize it. Because the surface defects have manifold shapes, we must take efforts to improve the quality of recognition and classification on surface defects pattern in steel strip. And to meet the requirement of detect all surface defects in the steel strip.