作者：孙荣荣,姜国金
作者单位：中国农业科学院柑桔研究所
摘要：农业生产决策是一个动态过程,必须依靠大量农业数据和强有力的数据分析技术作为支撑。特征提取技术能从大量数据中提取出反映源数据本质特点的特征,有利于农业数据的管理与分析。目前,地理信息系统已成为农业空间数据管理与分析的重要手段,科技工作者已从PCA和GIS相结合的角度进行了农业空间数据的特征提取研究,然而PCA所提取的特征不能够反映农业空间数据的非线性结构特征,所获取的特征不利于其后的评价与分析。因此,将非线性特征提取技术KPCA引入GIS中进行了农业空间数据特征提取,构建了农业空间数据特征提取模型。应用结果证明,通过该模型所提取的特征分布结果符合实际情况,为实现农业生产决策提供了更具保障的数据支持和科学依据。
关键词：核主成分分析,地理信息系统,特征提取,农业数据
Title: Research on the Feature Extraction of Agricultural Spatial Data Based on GIS and KPCA
Authors: Sun Rong-rong,Jiang Guo-jing
Abstract: The agriculture production decision is a dynamic science supported by agricultural data and analysis techniques.Feature extraction techniques can extract natural characteristically features from original data which advantage to the management and analysis of the agricultural data.On the other hand,GIS has widely obtained the approval as the method of management and analysis for agriculture spatial data.Researcher has been combined PCA with GIS on feature extraction of agricultural data,however,PCA is one of the nonlinear feature extraction method which can not deal with the nonlinear problems.A novel hybrid approach of GIS and KPCA for agricultural spatial data feature extraction was proposed.A model of feature extraction was crated and the application result showed that the feature extraction results matched the actual situation.
Key words: kernel principal component analysis; GIS; feature extraction; argricultural data
文献注录：孙荣荣,姜国金. 基于GIS和KPCA的农业空间数据特征提取研究 [J]. 农业网络信息. 2010, 02: 24-28.
基金: 重庆市烟草专卖局项目(2006016)
报/刊名：《农业网络信息》，发表于2010 / 第 02 期。
文献类型： [J]
页码：第 24-28 页 / 共5 页