基于单目视觉的水下目标三维信息重建方法研究

Research on 3D Information Reconstruction of Underwater Target Based on Monocular Vision

基于手眼协调原理,提出了一种采用单目摄像机对具有几何形状约束特征的水下目标进行三维信息重建的方法。方法首先对水下目标进行二值化处理,并对二值图像的几何信息提取与处理,然后配合机械手的动作获取目标的空间位置信息,进而通过视觉引导机械手与目标成特定的相对姿态,最后通过摄像机的空间位置与姿态信息推断出目标的空间位置与姿态信息。这种方法避免了传统机器视觉存在的水下目标模糊、纹理特征不明显、图像匹配困难等问题,根据二值图像的几何信息能够更稳定有效的获取水下目标的三维位置姿态信息。

Based on hand-eye coordination principle,a method of 3D information reconstruction of underwater target with geometric constraints using monocular camera is presented in this paper.The method extracts and analyses geometric information of underwater target after binary process firstly,and obtains the position of the target in collaboration with the action of the manipulator gradually,Then control the manipulator into a specific posture with target under the guidance of the vision,At the last the position and posture information of target can be infer from the position and posture information of camera.The method can avoid the problems that are emerged in traditional underwater machine vision field,such as fuzzy target image,unobvious texture,difficult image matching and so on.According to the geometric information of the binary images,this method can be more stable and effective to reconstruct 3D position and orientation information of the underwater target.