机载雷达测绘遥感图像下的典型空间信息数据挖掘方法  

Typical spatial information data mining methods under airborne radar mapping remote sensing images

作　　者：张桂莲 张东华 郑福海[4] ZHANG Guilian;ZHANG Donghua;ZHENG Fuhai(Administration of Surveying,Mapping and Geoinformation of Inner Mongolia,Hohhot 010010,China;Department of Land Science and Technology,China University of Geosciences(Beijing),Beijing 100083,China;College of Water Conservancy and Civil Engineering,Inner Mongolia Agricultural University,Hohhot 010010,China;The Second Geomatics Cartography Institute of Ministry of Natural Resources,Harbin 150081,China)

机构地区：[1]内蒙古自治区测绘地理信息中心,内蒙古自治区呼和浩特010010 [2]中国地质大学(北京)土地科学技术学院,北京100083 [3]内蒙古农业大学水利与土木建筑工程学院,内蒙古呼和浩特010010 [4]自然资源部第二地理信息制图院,黑龙江哈尔滨150081

出　　处：《湘潭大学学报（自然科学版）》2022年第4期60-70,共11页Journal of Xiangtan University(Natural Science Edition)

基　　金：国家自然科学基金(51969023);内蒙古自治区高等学校科学研究项目(NJZY20049);自治区重点研发和成果转化项目(2022YFDZ0050);内蒙古自治区科技计划项目(2020GG0076)。

摘　　要：机载雷达遥感图像测绘技术凭借诸多显著优势,为空间信息勘测精度带来新方式,但是机载激光雷达测绘遥感图像中的典型空间测绘信息,如地形信息、高程测绘角度等参数间的特征匹配度较低、数据可用性较低.设计一种典型空间信息数据挖掘方法.在机载激光雷达测绘遥感图像中,经对地定位、空间信息测距与扫描等一系列挖掘信息处理,采集空间信息数据,通过选取、预处理、架构训练数据集、模糊神经网络构建与训练等阶段,自适应调整模糊神经网络参数与结构,通过学习预处理后的训练数据来训练模糊神经网络,依据智能雷达坐标系与机载典型被测空间坐标系关系,获取空间信息数据的隐藏知识,实现空间信息数据挖掘.以机载雷达测绘遥感图像为基础设计对比实验,实验结果表明,所提方法控制点及外业调绘地物点的平面点位与高程的实测值与量测值偏差分别低于2.8 m、1.78 m和1.4 m、1.07 m,偏差较小;根据挖掘结果可知,廊道式扩展率在0.56%~1.82%之间波动,挖掘效果较好.综上可知,所提方法空间信息数据挖掘精度符合地理空间信息勘测需求,地域边界扩展情况分析结果较佳.Airborne radar remote sensing image mapping technology brings new methods for spatial information survey accuracy with many significant advantages.However,the typical spatial mapping information in airborne lidar remote sensing images,such as terrain information,elevation mapping angle and other parameters,has low feature matching and low data availability.Design a typical spatial information data mining method.In the airborne lidar mapping remote sensing image,after a series of mining information processing,such as ground positioning,spatial information ranging and scanning,spatial information data is collected.Through the stages of selection,preprocessing,building training data sets,fuzzy neural network construction and training,the parameters and structure of fuzzy neural network are adjusted adaptively,and the fuzzy neural network is trained by learning the preprocessed training data,According to the relationship between the intelligent radar coordinate system and the typical airborne measured spatial coordinate system,the hidden knowledge of spatial information data is obtained to realize spatial information data mining.In this paper,a comparative experiment is designed based on airborne radar mapping remote sensing images.The experimental results show that the deviation between the measured value and the measured value of the plane point position and elevation of the proposed method control points and field surveying and mapping feature points is less than 2.8 m,1.78 m,1.4 m and 1.07 m respectively,and the deviation is small;According to the mining results,the corridor expansion rate fluctuates between 0.56%and 1.82%,and the mining effect is good.To sum up,the accuracy of spatial information data mining of the proposed method meets the needs of geospatial information survey,and the analysis result of regional boundary expansion is better.

关 键 词：机载雷达 测绘图像 空间信息数据 数据挖掘 模糊神经网络 聚类算法 

分 类 号：P217[天文地球—测绘科学与技术]