带钢表面缺陷检测是带钢质量控制的重要环节之一，但现有带钢表面缺陷自动检测方法在准确性和实时性上还难以满足工业现场需要。为了解决此问题，提出了一种基于局部二进制模式（LBP）的带钢表面缺陷的初级检测方法。该方法首先利用快速局部二进制模式算法计算图像中各像素点的LBP值；然后通过统计LBP直方图来获取图像中主要边缘点的信息，再将其与阈值进行比较，以确定带钢图像中表面缺陷的存在，并记录缺陷的位置。实验结果表明，该方法不仅在带钢表面缺陷的初级检测方面具有良好的准确性和实时性，而且其提取出的信息还具有结构的和统计的双重特性，可为后续缺陷分类提供重要依据。

Strip surface detection is one of the basic process of strip quality control, existing methods for strip surface detection couldn’t meet the accuracy and real-time capability for industrial spots. To solve these problems, in this paper, a detection method of primary strip surface defect based on the local binary pattern (LBP) algorithm is proposed. Firstly, the LBP values of each pixel in strip image are calculated by employing a fast LBP algorithm. Then by constructing the LBP histogram, the information of principal edge points belonging to different types of defection is obtained. After thresholding, the existence and the location of defect in the image are suggested. Experimental results show, the proposed method not only has higher accuracy and real-time capability on the primary strip surface defect detection, but also can offer reliable structural and statistical feature information for further defect classification.