摘要
针对立体视觉障碍检测的两个难点:匹配精度以及匹配算法的实时性问题,文章提出了一种基于区域分割的障碍检测方法。通过区域分割,把图像对应点的匹配问题简化为区域边界点的匹配,提高了匹配的精度。分割后的区域作为一个独立的对象进行处理,障碍物三维信息计算转化为视差的判断,提高了匹配的速度。该算法能够实现在室内环境下的实时障碍检测,不需要利用环境的先验知识,具有较强的自适应性。实验结果表明了该方法的有效性。
A novel region-based obstacle detection algorithm for indoor navigation is presented in this paper.This method is different from the traditional ones,which solves two key problems in obstacle detection with stereo：matching accuracy and time consume.Firstly,the image is segmented into regions.Then every individual region is classified as belonging either to an obstacle or the ground based on its average ＇disparity＇.The method performs in real-time,and provides a local obstacle map at high resolution.Experimental results are shown for real indoor scene images and validate the efficiency of this method.
出处
《计算机工程与应用》
CSCD
北大核心
2005年第25期46-48,60,共4页
Computer Engineering and Applications
基金
国家自然科学基金(编号:60375037)
国家863高技术研究发展计划(编号:2003AA421030)
关键词
自主式移动机器人
障碍检测
逆投影变换
区域分割
autonomous mobile robot,obstacle detection,inverse perspective mapping,region segmentation
