研究了自主式水下机器人(AUV)利用单目光视觉系统对水下管道的检测跟踪问题,对实现该方法的各个过程从硬件组成和软件体系结构两方面进行了阐述,从而完成了一整套水下光视觉系统的软、硬件设计.分析了水下管道图像的特点,结合该特点阐述了图像预处理过程,提出了一种改进的Hough变换方法,改善了管道边界的提取效果.同时采用管道参考区域预测方法,提高了管道的检测率以及实时性,并利用水池试验对系统的可行性和有效性进行了验证.
The detection and tracking of underwater pipeline based on monocular vision system of AUV(autonomous underwater vehicle) were addressed.An integrated vision system including software and hardware structure was designed.The feature of underwater pipeline image was analyzed and according to that,corresponding pre-processing method was proposed.An improved Hough transforming was advanced to obtain pipeline contour.Pipeline reference zone was estimated to increase the accuracy and decrease the time cost of pipeline tracking.Finally,the system was validated to be effective and feasible through the tank experiments.