摘要:
地表覆盖类别数据尺度上推方法中仍以众数选择或比例占优为原则,但这2种原则都存在一定的缺陷,比如在多个众数的情况下随机选取会使地图信息很大程度上失真.为了更好地保持尺度综合前后地物类别空间信息,将地表覆盖类别优先准则和语义邻近规则引入到尺度上推方法中,在众数聚合规则上考虑类别优先级和语义邻近度,建立一种新的地表覆盖类别数据升尺度方法.该方法通过设定类别优先级和语义邻近度,计算得到粗尺度上对应空间位置的类别,从而完成地表覆盖类别数据的尺度综合.实验结果表明,在众数聚合方法基础上,加人类别优先级和语义邻近度,能够较好地保持尺度综合前后地物空间分布的一致性,地图相似度也相对较高.
Abstract:
The upscaling methods for land cover classification data are still based on mode number selection or principle proportion choice;nevertheless,blocked at multiple modes,map information will be largely distorted under randomly choice.In order to maintain spatial information before and after generalization,this paper introduces land cover priority guidelines and semantic proximity norms to the upscaling method.A new upscaling method can be obtained by adding both rules to the aggregation processing.Firstly,through setting land cover priority guidelines and semantic proximity norms,the output category at the corresponding spatial location can be calculated on the coarse scale.The results show that,based on the majority aggregation method,and by adding the priority of the neighboring and semantic rules,the consistency of the spatial distribution of the feature can be better maintained after generalization,and the similarity of original maps is relatively high.
下载: