水下目标三维视觉信息复原是水下智能机器人（AUV）感知环境信息、进行自主作业的关键技术。本文介绍了一种基于模型的复原方法，即通过检测目标边缘拟合轮廓，根据模型知识复原三维信息。该方法已经成功地导引AUV进行了水下自主避障、自主剪缆、自主投掷等智能作业。

The restoration of 3-D vision information about underwater targets is the key problem for an Autonomous Under water Vehical (AUV) to sense envirronmental information and perform autonomous tasks. This paper presents a model-based restoration method. This method makes the curve fitting based on the detected edge points of targets and restores the .3D information of targets using the model knowledge. The method has been employed, successfully, for the guidance of an AUV in its autonomous underwater operations.