论文部分内容阅读
花纹板是众多钢材产品中的一种,由于其复杂的表面特性,导致现有的带钢表面质量在线检测系统无法布置。尝试通过对花纹板中一种最常见的伪缺陷——表面水痕,进行分析研究,给出一种在包含有纹理背景的情况下的缺陷检出方法,为在花纹板上布置带钢表面质量在线检测系统提供有效支撑。通过对图像像素灰度特性、梯度特性以及相互之间关联度特征进行分析,给出一种基于游长编码思想的花纹板表面水痕标记算法,通过该算法的应用,能够准确高效地定位图像上的水痕,为将水痕伪缺陷排除出花纹板缺陷提供技术保障。
The checkered plate is one of a large number of steel products which, due to its complex surface characteristics, leads to the inability of the existing on-line inspection system of strip surface quality to be arranged. Attempt to analyze and study the watermark, one of the most common pseudo-defects in the checkered plate, gives a method of detecting the defects in the case of containing the textured background, in order to arrange the strip surface Quality online testing system to provide effective support. By analyzing the features of grayscale, gradient and inter-relatedness of image pixels, an algorithm of watermark marking on the surface of the checkered plate is given based on the idea of long-length coding. By using this algorithm, the image can be accurately and efficiently located On the water marks, for the water marks false defects excluded from the checkered plate defects provide technical support.