基于二维类间方差多门限阈值分割的石块识别方法  

Stone Detection by Approach of MultiThreshold Segmentation Based on Between-Cluster Variance

作　　者：王荣本[1] 顾柏园[1] 郭烈[1] 余天洪[1] 

机构地区：[1]吉林大学交通学院,长春130025

出　　处：《计算机工程与应用》2006年第28期230-232,共3页Computer Engineering and Applications

摘　　要：月球环境感知是未来月球车在月球表面自主导航的关键。石块作为月球表面的一种地理特征,将成为月球车行驶的障碍物。文章从模式识别的分类理论出发,提出了改进的二维类间方差的多门限阈值石块分割方法。在获取分割图像后,利用石块底部阴影信息和原始图像的灰度直方图分布特征,将分割图像二值化。然后,结合形态学滤波算法和二值图像投影变换,实现了对部分场景下的石块识别和定位。Moon environment exploration is very important to moon vehicle autonomous navigation on moon surface. Stone as a kind feature of moon surface,it will be a kind of obstacle to moon vehicle run.In the paper,firstly,using the classification theory in pattern recognition,approach of muhithreshold stone image segmentation based on improved 2D class variance is presented.Second, Binary image is obtained by utilize the information of shadow of stone and the gray image histogram.Then,using morphological filter and projection transform,detection and position of stone is utilized in part environment.

关 键 词：月球环境 石块识别 多门限阈值 类间方差 

分 类 号：U495[交通运输工程—交通运输规划与管理]