动态环境下四腿机器人图像处理的改进  

DImage Processing Improvement in Dynamic Environment Based on Four-legged Robot

作　　者：张枨[1] 许东来[1] 余跃庆[1] 

机构地区：[1]北京工业大学机电学院

出　　处：《微计算机信息》2009年第5期211-212,191,共3页Control & Automation

基　　金：基金申请人:许东来;项目名称:含有被动关节的柔性机械臂动力学机器控制的理论与实验研究;基金颁发部门:国家基金委(50575002)

摘　　要：视觉系统是自主体机器人感知外界坏境信息的重要环节。颜色分割是视觉系统中图像处理的第一步。目前公认效果较好的方法采取的是颜色表查找法。本文以四腿机器人足球比赛为背景和研究平台,提出一种快速制作颜色表的方法,并在实时环境下运用了图像增强技术,改进了颜色分割的算法,提高了机器人视觉系统的鲁棒性,并应用于2006年的RoboCup中国机器人大赛中,得到了较好的效果。Vision system is the key system for agent robot to sense its outside environment. In vision system, the first step to process images is the color segmentation. Nowadays, it is generally acknowledged that color-table-lookup works very well. On the back-ground and research platform of four-legged robot soccer competition, this article provides a method to improve the manually created color table. Meanwhile, it improves the,algorithm of the color segmentation using image enhancement under the real-time circumstance, which makes the robot vision system reach a high degree of robustness. Meanwhile, it was applied in RoboCup China Open 2006 and proved to be an effective way.

关 键 词：机器人视觉系统 图像处理 鲁棒性 颜色表 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]