摘要： 当机器人载体摄像机观察目标为某些几何特征缺乏，仅有纹理特征场景时，基于几何特征视觉计算的伺服手段难以应用，为此该文提出基于纹理的三自由度视觉伺服方法。分析摄像机运动与目标图像纹理粗糙度特征变化之间的关系，给出基于纹理粗糙度的视觉伺服方法。通过机器人伺服悬停半物理仿真实验证明，分析纹理图像特征变化可以提供完成伺服任务所需要的运动信息。

关键词: 计算机视觉, 位姿保持, 视觉伺服, 纹理, 特征提取

Abstract: This paper proposes a three DOF vision servoing method based on texture analyzing. The theory based on geography characteristics calculation will not be effective in the environment full of texture but short of geography characteristics. By analyzing the relationship between the camera motions and the coarseness changing of the texture image, a vision servoing method based on texture coarseness analyzing is given. The result of half-physical robot station keeping experiment shows that the changing of texture characteristics can provide enough motion information for the servoing task of the robot.

Key words: computer vision, station keeping, vision servoing, texture, characteristic extraction

中图分类号: 

• TP391