无人驾驶车野外环境水体障碍物检测  

Water obstacle detection in off-road environment for unmanned ground vehicles

作　　者：赵一兵[1,2] 邓云翔 郭烈[1,2] 潘驰[3] 

机构地区：[1]大连理工大学汽车工程学院,辽宁大连116024 [2]工业装备结构分析国家重点实验室,辽宁大连116024 [3]大连交通大学管理学院,辽宁大连116028

出　　处：《中国科技论文》2013年第11期1192-1196,共5页China Sciencepaper

基　　金：高等学校博士学科点专项科研基金资助项目(20110041120024);国家自然科学基金资助项目(51205038);中央高校基本科研业务费专项资金资助项目(DUT13JS14;DUT13JS02)

摘　　要：野外环境下水体障碍物会对陷入其中的无人驾驶车造成损害。针对静态水面特征呈现出的较高亮度、较低饱和度以及平滑的纹理特征,提出将饱和度亮度比值颜色特征和从灰度共生矩阵中提取的纹理特征融合的野外水体障碍物检测方法。对图像进行HSV颜色空间转换提取颜色特征,计算灰度共生矩阵提取纹理特征,构成特征矩阵并训练SVM分类器用于实际检测。仿真和试验结果表明该水体障碍物检测方法有效,并能够有效地减弱光照变化对单纯基于颜色特征检测水体的负影响。Water hazard in off-road environment may destroy electronic components of unmanned ground vehicles. Based on the features of high intensity, low saturation and low texture of the water hodies existing in off-road environment, we propose a method combining saturation-value ratio color features with texture features extracted from gray level co-occurrence matrix （GL- CM）. Images are converted into HSV color space for color feature extraction. GLCM is calculated for texture feature extraction. Feature matrix is used to construct SVM classifier which would be tested by image set. Experimental results demonstrate signifi- cant progress on detection of water body hazard in off-road environment, which effectively reduces the influence of illumination variation on detection when only using color feature to detect.

关 键 词：计算机图像处理 HSV颜色空间 纹理特征 灰度共生矩阵 SVM支持向量机 

分 类 号：U469[机械工程—车辆工程] U467[交通运输工程—载运工具运用工程]