基于机载激光雷达数据识别单株木的新方法  

A New Method of Individual Tree Recognition based on Airborne LiDAR Data

作　　者：唐菲菲[1,2] 阮志敏[3] 刘星[1,2] 张亚利[1,2] 

机构地区：[1]重庆大学土木工程学院,重庆400044 [2]山地城镇建设与新技术教育部重点实验室(重庆大学),重庆400045 [3]招商局重庆交通科研设计院有限公司,重庆400067

出　　处：《遥感技术与应用》2011年第2期196-201,共6页Remote Sensing Technology and Application

摘　　要：在林业应用中,机载激光雷达技术较被动式遥感技术有着独特优势。以从机载激光雷达数据中分离单株木,提取单株木树高和树冠大小信息为目的,融合有标记约束的分水岭分割和流域跟踪分割两种图像分割方法,提出一种新的单株木识别思路。以美国某地区实地采集激光雷达数据为例验证本文提出方法,实验结果表明该方法通过增加边缘检测范围的约束条件,能够有效避免过分割现象,并通过使用约束条件,减少在其检测范围内的目标数量,从而避免不必要的检测干扰,较传统方法能快速准确地识别单株木。In forestry applications,Airborne LiDAR has its unique superiorities compared to passive remote sensing.In this paper,a new individual tree recognition method was proposed to isolate individual tree and extract tree height,also the area of the tree crown,this method is the fusion of marker-controlled watershed segmentation and valley-following segmentation.Finally,a real LiDAR dataset was test,experiment results show that this method can perform edge detection effectively compared to methods before because detection constraint was added into it.The detection constraint can avoid over-segmentation to some extend,and reduce the detection objects,which can avoid unnecessary detection noise.

关 键 词：机载激光雷达 单株木 树高 树冠 图像分割 标记 

分 类 号：P225.7[天文地球—大地测量学与测量工程]