摘要
针对单目摄像头工作的双足机器人的视觉定位问题,设计了一种视觉传感器与超声波传感器相结合的信息融合方法,进行目标的识别和定位;对拍摄的图像进行HSV颜色空间的阈值分割得到二值图像,并对其进行形态学滤波处理和Hu轮廓不变矩匹配,以此识别到目标和其重心点图像坐标;建立单目定位模型,用机器人自身的超声波传感器得到与目标物之间的距离,采用Zhang标定法得到摄像机参数,把这些已知数据带入坐标转换,得到目标物在世界坐标系下的坐标,实现定位;借助NAO机器人的平台,将基于信息融合的定位方法和一般单帧图像定位方法进行对比实验;实验结果表明上述方法的误差相对较小,且能得到目标物的三维信息,验证了该定位方法的实际可行性。
In order to realize recognition and orientation of target, an information fusion method of visual sensor and sonar sensor is pro posed. Taking threshold segmentation based on HSV color space, morphological filtering and Hu contour moment invariants to identify target and get its image coordinate of center of gravity point. The Monocular Vision Location Model is established, the distance between the tar get and the robot is obtained by ultrasonic sensor, the parameters of camera are calibrated by employing Zhang＇ s algorithm. The coordinates of the object in the world coordinate system are obtained by coordinate transformation using the data from the steps above, to realize the localization. Therefore, with the platform of NAO robot, presenting a comparative experiment of the localization methods based on information fusion and the ones based on normal single frame image. The experimental results show that the error of the proposed method is relatively small, and the three--dimensional information of the object is accessible, verifying the practical feasibility of the proposed localization method.
出处
《计算机测量与控制》
2016年第11期192-195,200,共5页
Computer Measurement &Control
关键词
信息融合
识别与定位
图像处理
NAO机器人
information fusion
recognition and loealization
image processing
NAO robot
