针对基期(用于该研究的前一时期数据)T1仅拥有土地利用和覆盖图件(矢量格式)而另一期T2拥有遥感数据的情况,构建了基于知识引导的土地利用和覆盖变化自动检测技术与方法。T1时期土地利用与覆盖与T2期遥感数据在配准叠加情况下,以T1完整的土地利用与覆盖类型图斑为单元构建土地各类别遥感数据知识库,然后以图斑单元或以像素为单位计算遥感影像特征统计量,通过与知识库相关数据的比较与匹配自动检测出变化并识别出相应的土地利用与覆盖类别。文章最后通过试验验证了该方法的有效性。

Automatic land use and land cover change (LUCC) detectionvia remote sensing image has awide application inthe\narea of LUCCresearch, nature resource and environmentmonitoring and protection. Underthe conditionthatone time (T1) data\nis existed land use and land covermap, and anothertime (T2) data is remotely sensed imagery, howto detect change automati-\ncally is still an unresolved issue.\nThis paper developed a land use and land cover classknowledge-orientedmethod forautomatic change detectionunderthis\nsituation. Firstly, the land use and land covermap inT1and remote sensing images inT2were registered and superimposed pre-\ncisely. Secondly, the remote sensing knowledge database of all land use and land cover classeswas constructed based onthe un-\nchanged parcels inT1map. Thirdly, guided byT1land use and land covermap, feature statistics for each parcel or pixel in re-\nmote sensing imageswere extracted. Finally, land use and land cover changeswere found and the change class was recognized\nthrough the automatic matching between the knowledge database of remote sensing information of land use and land cover classes\nwith the extracted statistical information inthat parcel or pixel. Experimental results and some practicat applications showthe ef-\nficiency of this method.