中文摘要：

在高速、宽幅的纸病视觉表面缺陷检测系统中,必须快速有效的检测出可疑目标,以满足大数据量的实时检测要求.在分析传统的差影减法实现目标检测算法的的基础上,基于灰度数学形态学原理,采用Top-Hat运算去除图像噪声并完成纸病图像的可疑目标的快速检测.实验表明,基于图像灰度的目标检测算法快速、健壮,抗噪性能好,实现了可疑目标的快速检出.

英文摘要：

In order to meet the requirements of high data throughput and real-time inspection, fast candidate object detection algorithm must be adopted in high speed and wide breadth vision inspection system for surface defects image. After analyzing traditional image difference subtraction algorithm, based on the theory of mathematical morphology, Top-Hat operation algorithm is chosen to remove image noise and perform the task of object detection based on gray morphology theory. The result of experiments shows that the new algorithm is fast, robust and noise proof and candidate object is quickly detected.