Abstract：This paper discusses how to recognize and locate pipeline accurately by vision in underwater environment. An underwater vision system is presented. The object′s color is the feature when the object be recognized. And the color is transformed from RGB space to HSI space, which improves speed of image processing and robustness of feature extraction. Because the illumination of underwater environment changes tempestuously, fuzzy network is used to adjust threshold value of feature color dynamically. The paper analyzes location of point and line in underwater environment in detail, and offers an algorithm of location. Experiment results show that the method can achieve quite precise location effectiveness.
 陈荣盛,袁小海,胡震等. 水下机器人目标搜索与识别算法. 中国海洋平台,1999,14(4)  Castleman K R. Digital Image Processing. 北京:清华大学出版社,1998,9  周煦艟,施鹏飞.快速HLS彩色空间变换方法.上海交通大学学报,1998,32(9)  IiHong Suh,Tae Wonkim. Fuzzy Membership Function Based Neural Networks with Applications to the Visual Serving of Robot Manipulators. IEEE Transactions on Fuzzy Systems, 1994,2(3)  赵振宇,徐用懋.模糊理论和神经网络的基础与应用.北京:清华大学出版社,1996  Shakunaga T. An Object Pose Estimation System Using a Single Camera. IEEE Int Conf Intelligent Robots and Systems,1992:1053-1060  钟玉琢.机器人视觉技术.北京:国防工业出版社,1994,11  Horn B K P. Robot Vision. USA:The MIT Press,1986