基于分类估计的异质数据融合M_(split)模型  

M_(split) Model for Fusion of Heterogeneous Data Based on Classification Estimation

作　　者：陶叶青[1] 束明聪 陈浩 TAO Yeqing;SHU Mingcong;CHEN Hao(School of Geography and Planning,Huaiyin Normal University,Huai'an 223300,China)

机构地区：[1]淮阴师范学院地理科学与规划学院,江苏淮安223300

出　　处：《淮阴师范学院学报(自然科学版)》2025年第1期45-51,共7页Journal of Huaiyin Teachers College(Natural Science Edition)

基　　金：安徽省高校自然科学研究项目(2022AH040297);芜湖职业技术学院优秀青年拔尖人才项目(rc2023qnbj08);芜湖市应用基础研究项目(2022jc47);芜湖职业技术学院创新团队项目(wzykytd202201);安徽高校质量工程研究项目(2022jnds064)。

摘　　要：从分类估计的角度研究异质数据融合理论,建立一种不依赖于随机模型的融合方法.该理论首先基于M_(split)估计对观测数据进行分类估计,然后根据分类估计结果,应用中位函数计算不同参数估值的中误差,最后通过定义尺度因子与相对权比函数构建融合模型,实现异质数据的有效融合.通过两类不同精度量级的观测数据组成的控制测量实例对该方法进行验证,结果表明,该方法易于实现,得到的参数估值稳定.This paper investigates theory for the fusion of heterogeneous data from the perspective of classification estimation,and establishes a fusion method which does not depend on stochastic model.This theory utilizes Msplit estimation to classify observation data,and employs median function to calculate mean square errors of different parameters estimations,according to classification estimation results.Finally,fusion model is established by definitions of scale factor and relative weight radio function to realize the fusion of heterogeneous data effectively.An instance of control survey which is composed of two categories of observation data belonging to different accuracy magnitudes is employed to verify feasibility of the method presented in this paper.The results demonstrate that this method is easier to realize,and the parameters estimation are stable.

关 键 词：异质数据融合 分类估计 M_(split)估计 尺度因子 融合模型 

分 类 号：P207[天文地球—测绘科学与技术]