中文摘要：

热轧带钢表面的温度高,辐射光强,并且存在着水、氧化铁皮、光照不均等现象,与冷轧带钢的表面存在着很大的差别。将线阵CCD摄像机作为图像采集装置,用绿色激光线光源作照明,通过窄带滤色镜滤除钢板表面的辐射光,从而提高了缺陷对比度。根据热轧带钢表面的特点提出新的缺陷检测与识别算法流程,通过增加4种不同类型的缺陷检测步骤,去除了大量由水、氧化铁皮等造成的伪缺陷,在保证缺陷检出率的同时,减小缺陷的误识率。经在线应用,该方法可以满足在线检测的要求,缺陷的检出率达95%以上,识别率达85%以上。

英文摘要：

Surface of hot-rolled steel strips is different from surface of cold-rolled steel strips, because it has high-temperature and intense radiation. Moreover, there are waters, scales, uneven illumination on hot-rolled steel strips. Line-scan CCD cameras are introduced to capture images, and green linear lasers are applied for illumination. Radiation of hot-rolled steel strips is filtered by narrow-band filter lens, and contrast of defects is improved. According to the characteristics of hot-rolled steel strips, a new algorithm of defect detection and classification is developed. A step of defect detection for four different types of detects is added to remove false defects caused by waters, scales, etc., and false detection is decreased while detection rate of defects is not worsen. Results of industrial application show that the technology can be applied to on-line inspection, and detection rate of defects is higher than 95% while classification rate of defects is higher than 85%.