车载单线激光雷达成像分割及噪声检测算法  

Vehicle-based single-line laser radar image segmentation and noise detection algorithm

作　　者：梁雄[1] 唐琎[1] 蔡自兴[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083

出　　处：《计算机应用研究》2011年第5期1975-1977,共3页Application Research of Computers

基　　金：国家自然科学重大专项基金资助项目(90820302)

摘　　要：针对国内外已有的激光雷达2D成像分割聚类算法,提出了基于ABD(adaptive breakpoint detector)算法检测分割点后的再次分割聚类判定的方法,通过实验对比可以发现,改进后的算法使分割效果更加理想。针对激光成像中的噪声数据,提出了基于上述改进分割算法的有效简易检测算法,噪声点检测效果理想。针对激光雷达可能会扫描到路面数据,但该路面数据事实上不属于障碍物数据,提出了基于当前帧和前一帧局部区域及整个区域数据突变相结合的新型检测算法,实验结果表明路面检测效果理想。上述研究内容对于车载传感器数据融合具有一定价值。After analyzed laser radar image segmentation algorithm at home and abroad,this paper proposed an improved algorithm based on ABD.The experiments show that the new segmentation algorithm is more rational and effective.And also proposed an efficient method for noise point detection based on the improved segmentation algorithm,the results were ideal.The single-line laser radar would detect the road surface,but actually it was not a real obstacle,proposed a local and entire region combining algorithm based on the tremendous difference of current frame and previous frame.Experiment results show that the road detection is effective.The research mentioned above is valuable for vehicle sensor data fusion.

关 键 词：激光雷达 2D成像 分割聚类 噪声点检测 路面检测 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]