摘要
水下机器人-机械手系统(UVMS)在对水下目标进行自主或半自主作业时,首要及关键的一步是进行目标识别。由于水下环境的特殊性,水下视距很小,光学成像具有噪声大、对比度差等特点;上述因素使得基于局部特征的目标识别对尺度、旋转、光线等变化的敏感度大大增加。借助立体视觉及航向传感器信息,利用水下三维目标的变尺度形态图方法可靠地实现了水下目标识别,并通过实验对该方法进行了验证。
For a UVMS,it is a prime and critical step to recognize the target while conducting autonomouse or semi-autonomous operations.Due to the particularity of underwater imaging,the visibility is quite limited and the images usually have much noise and low contrast.Such factors increases the sensitivity of the recognition based upon local features,to scale,rotation or illumination.With stereo measurement and heading information,a scalable aspect graph method is used to enhance the reliability of underwater target recognition,which is also validated through experiments.
出处
《机械设计与制造》
北大核心
2011年第3期265-267,共3页
Machinery Design & Manufacture
基金
国家"863"资助项目(2006AA09Z217)
关键词
水下机器人-机械手系统
形态图
立体视觉
目标识别
Underwater vehicle-manipulator system
Aspect graph
Stereo vision
Target recognition
