基于自主车导航的障碍物分割方法  

Obstacle segmentation approach in ALV navigation

作　　者：夏永泉[1] 黄敏[1] 郭龙源[2] 杨静宇[2] 

机构地区：[1]郑州轻工业学院,河南郑州450002 [2]南京理工大学,江苏南京210094

出　　处：《计算机工程与设计》2009年第7期1810-1812,共3页Computer Engineering and Design

基　　金：国家科技支撑计划基金项目(2006BAK01A38);郑州轻工业学院博士基金项目(2007BSJJ005);国防基础研究基金项目(J1500C002)

摘　　要：对目标进行有效的分割是障碍物检测中重要的一个环节,提出了基于特征区域的分割方法。该方法通过定义特征函数将整个图像分为特征像素集合和非特征像素集合,前者包含了边缘等特征信息,后者包含了平滑区域像素;对将两个集合内各像素值分别置为1和0得到二值图像,对二值图像进行外轮廓跟踪得到最后的分割结果。实验结果表明该方法是有效可行的。Segmentation is the important step in obstacle detection. A simple segmentation approach based on character function is pro- posed. First, the approach divides all the pixels of image to two group through defined characteristic function, one is characteristic pixel group which includes character information, such as, edges, contours, and so on, the other is non-characteristic pixel group which includes pixels in smooth region. The pixels in two group are set to 1 or 0 respectively. As a result, original image is characterized to binary image. After that, extra-contour tracing approach is employed to trace the extra contour of the characteristic region, the tracing result is the last segmentation result. The experiment result indicates that the proposed segmentation is feasible and valid.

关 键 词：障碍物检测 特征区域 特征函数 二值图像 轮廓跟踪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]