基于图像矩的眼注视研究

Gazing Based on Image Moments

利用图像的全局特征描述子—图像矩特征作为图像特征信息 ,从而构成视觉伺服 ,实现对动目标的注视跟踪 .俯仰由末端执行器完成 .摇摆由腰的旋转完成 .最后给出了实验结果 ,表明了矩特征的有效性以及五自由度机器人实现注视跟踪的可能性 .

This paper describes using image moments which are the global image descriptors as image features for visual servoing.Object tracking is accomplished by gazing.The robot end effector pitches and the waist rotates instead of jawing of the end effector.Finally experiments are provided to show that image moments as image features for visual servoing are effective and it is possible to gaze the dynamic object even on a 5 degrees of freedom robot.