考虑空间相关性的数据空间特征提取方法  

A Spatial Feature Extraction Method Considering Spatial Correlation of Data

作　　者：肖启华[1] 陈珂[1] 黄冬梅[1] 

机构地区：[1]上海海洋大学信息学院,上海201306

出　　处：《计算机仿真》2014年第12期425-428,433,共5页Computer Simulation

基　　金：国家自然科学基金项目(61272098);上海海洋大学首届人才计划(海燕计划:B-5003-11-0056)

摘　　要：针对具有空间分布特性的时空数据,首先考虑数据的空间相关性对数据进行空间系统抽样,形成灾害数据空间分布样本;然后考虑样本数据的时间维度,运用粗糙集数据约简方法提取数据的空间特征。空间抽样方法能保证抽样精度、提高抽样效率;基于样本数据的约简技术能在保证数据主要传达信息不丢失的前提下对数据进行有效压缩、极大降低数据量,是解决海洋灾害数据监测花费巨大、数据分析繁杂的一种客观方法。最后,以某浅滩某次海洋灾害时空数据进行仿真,验证了数据简约方法的优良性。According to the spatial distribution characteristics, the disaster data can be sampled with spatial sys- tematic sampling considering the spatial relevance of data firstly, to form the disaster sample data. Secondly, spatial features were extracted with data reduction teleology based on rough set from the disaster sample data considering the time dimension of data. Spatial sampling ensures the efficiency and precision of sampling. And the data reduction technology based on sampling can not only retain the main information of disaster data, but compress the data effectu- ally and greatly reducing the data quantity, so combination of the two methods forms an new objective method that can solve image quality problems such as non - fluency and in coherence in visualization of disaster with large data. Computer simulation based on some disaster of a shoal verified the good property of the method mentioned in paper.

关 键 词：时空数据 空间特征 空间相关性 抽样 数据约简 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]