基于OpenCV的麦克纳姆轮全向平台的视觉循线设计

摘要

针对仓库、厂房等地对运送平台自主移动的需求,文中着重进行了麦克纳姆轮全向移动平台的摄像头循线算法优化设计.通过HSV与RGB结合分割的方式适应多背景的颜色分割.采用概率霍夫变换检测直线后,提出一种简化算法,可准确获得目标直线并减少运算量.在拐角与十字路口检测时引入凸包缺陷检测,通过算法的优化与几何条件的综合判定,做到准确检测.采用OpenCV编程,移植到Manifold嵌入式处理平台后,经实际测试证明,该算法能够达到实时分析的要求.%In order to satisfy the needs of the autonomous mobile transport platform in warehouse and factory,patrol line by camera of omni-directional platform with me can um wheel is studied in this article.Combination of HSV and RGB color segment is used to accommodate different kinds of background.Probabilistic hough transform is used to detect line,and an algorithm is proposed to get target line information and reduce the operating quantity.Convexity defects is used to detect turn and crossroad.The accuracy of detection is improved by the optimization algorithm and comprehensive judgment of geometry conditions.The program is based on OpenCV.It is transplanted to Manifold embedded platform.The actual test proved that the algorithm can achieve real-time analysis requirements.