摘要
平面标靶的定位是3维激光扫描数据处理过程中的重要一环。为了减小中心坐标提取对多站数据拼接配准、不同坐标系转换、辅助定位等精度的影响,提出了一种基于最小二乘法、K-means聚类法以及扫描点反射强度值的新方法,并进行了理论分析和实例验证,取得了准确的定位数据。首先采用总体最小二乘法对扫描数据中的粗差点进行剔除,然后基于扫描点反射强度值和K-means聚类方法进行区域分类,接着对中心坐标提取,最后对中心坐标进行计算得到坐标位置。结果表明,在一定距离内,该方法可以在实际应用时获得较高的定位准确性,在不同方向上与软件自动识别的坐标相差在亚毫米级,与实际测得标靶距离相差也在亚毫米级。该研究对实现平面标靶中心坐标自动提取是有帮助的。
Plane target positioning is one of the important parts to process the data obtained by terrestrial laser scanner measurement. In order to reduce the influence on the accuracy of multiple-scan registration, coordinate system transformation and object aided location when extracting coordinates of the target centre,a new method was proposed based on the least square method,K-means clustering method and the reflection intensity at the scanning points. After theoretical analysis and experimental verification,the positioning data with high accuracy was achieved. There are four major steps to deal with the problem： firstly,to eliminate noise points by using the total least squares algorithm; secondly,to classify the region of scanning data based on the reflection intensity; thirdly,to extract the centre coordinates of each region by means of K-means clustering method; finally,to calculate the target centre coordinates according to the centre position each region. The results show that reliable positioning accuracy can be obtained at a certain distance by means of the proposed method. The deviation of the coordinates is in the order of sub-millimeters,and the distance deviation is also in the order of sub-millimeters. The research is helpful for coordinate extraction of plane target centers.
出处
《激光技术》
CAS
CSCD
北大核心
2015年第3期341-343,共3页
Laser Technology
基金
中央高校基本科研业务费资助项目(DUT14QY42)
2013年大连市建设科技发展计划资助项目
作者简介
吴超（1987-），男，博士研究生，主要从事自动化数据采集技术在工程中应用方面的研究。
通讯联系人。E-mail：yongbo@dlut．edu．cn
