摘要
规则物体的识别与抓取往往比非规则物体的识别与抓取更加有规律性,因此本文以规则的圆形、长方形物体为例,对圆提出圆形度这种度量方法,对长方形基于角点检测,提出计算各个角点之间夹角,结合HSV模型对图像进行分割,提高物体识别度。为了使单目机器人更准确的抓取物体,需要比较准确的知道目标物体相对于机器人的位置信息,结合地平面约束测距模型算法计算圆形物体的具体世界坐标。在Visual Studio2010,Pioneer3-AT机器人平台上实验表明,基于圆形度的物体识别方法能够比较快速准确的识别圆形物体,进而准确抓取。
The recognition and grasping of rule objects are more regular than those of irregular objects.Therefore, this paper takes rectangular circular and objects as examples.For example,the author puts forward circularity this metric to recognition circle object,rectangle based on corner detection is proposed to calculate the angle between each corner of it,by combining HSV model for image segmentation.In order to grasp object more accurately,the author uses monocular robot objects,needs more accurate knowing the relative location information between circular object and the robot's.Therefore,this paper combines with ground plane constraint ranging model algorithm to calculate the specific world of circular objects.The experimental results show that through Visual Studio2010 and Pioneer3-AT robot platform show that the method can quickly and accurately identify the circular object,thus grasping accurately.
出处
《软件》
2016年第2期89-92,97,共5页
Software
