为了实现对系留绳外护套表面损伤的实时在线检测,设计了一套基于机器视觉的自动检测系统。根据文中采集图 像的特点,提出了改进的背景校正方法,同时运用自适应投票的快速中值滤波方法消除噪声,利用统计阈值分割法进行损 伤检测并二值化,经形态学处理剔除伪损伤,并获取损伤边界,最后通过提取损伤的形态特征实现分类识别。实验结果表 明,该系统的图像处理算法简单有效,检测速度快,检出率高,能够满足实际应用。

In order to detect defects on the surface of outer jacket of mooring cable real-time and online, an automatic detection system is designed. In this paper, according to acquired images' characteristic, a novel background correction method is proposed. Base on its result, an auto-adapted voting fast median filtering algorithm is used to eliminate noises, and statistical threshold method is used to segment defects, then the image is processed by morphology method to eliminate fake defects and obtain their boundaries. Finally, distilling defects' morphological features to realize classification and recognition. Experiment result shows that those image processing methods in the system are easy, effective, and with high detection speed and rate, so the system can satisfy practical application.