基于字典学习的土壤地球化学数据稀疏重构  

Pedogeochemistry data sparse reconstruction based on dictionary learning

作　　者：吴琳 魏友华[1] 洪姗 WU Lin;WEI Youhua;HONG Shan(Geomathematics Key Laboratory of Sichuan Province,Chengdu University of Technology,Chengdu 610059,China)

机构地区：[1]成都理工大学数学地质四川省重点实验室,四川成都610059

出　　处：《现代电子技术》2019年第9期18-21,共4页Modern Electronics Technique

基　　金：国家重点研发计划(2017YFC0601505);国家自然科学基金面上项目(41672325);四川省科技厅应用基础项目(2017JY0209)~~

摘　　要：在土壤地球化学数据的勘察和采集过程中,因为各种客观因素导致数据不完整,这会对后续的研究工作造成一定的影响,因此对数据进行重构是预处理阶段最基本的步骤。基于土壤地球化学数据在自身或在变换域内的稀疏性,建立基于字典学习的土壤地球化学数据重构模型,将数据重构问题转化为稀疏优化问题,可以减少数据重构后的平滑效果,在一定程度上保留土壤地球化学数据在异常区和背景区交界处的结构特征。最后将反距离插值法和稀疏重构算法重构后的数据进行对比,结果表明稀疏重构算法能有效地对土壤地球化学数据进行重构。In the process of the exploration and collection of pedogeochemistry data,the data is incomplete due to various objective factors,which has the influence on the subsequent research to some extent. Therefore,the reconstruction of the data is the most fundamental step of pretreatment stage. On the basis of the sparsity of the pedogeochemistry data itself or in the transform domain,a pedogeochemistry data reconstruction model based on dictionary learning is established,which can convert the data reconstruction problem into sparse optimization problem,so as to reduce the smooth of reconstructed data,and reserve the structure feature of pedogeochemistry data at the junction of the abnormal area and background area to a certain extent. The data reconstructed by inverse distance interpolation method and sparse reconstruction algorithm is compared,which shows that the sparse reconstruction algorithm can reconstruct the pedogeochemistry data effectively.

关 键 词：土壤地球化学 稀疏优化 数据重构 字典学习 正交匹配追踪算法 反距离插值法 

分 类 号：TN911.1-34[电子电信—通信与信息系统]