基于模糊子集的土地利用遥感图像模糊规则分类

Land Use Classification of Remote Sensing Images Fuzzy Rules based on Fuzzy Subsets

为了较好地处理遥感图像的不确定性或模糊性,提高分类精度,提出了一种基于模糊子集的土地利用遥感图像模糊规则分类方法。将模糊隶属度函数值对应到特定的模糊子集建立模糊规则条件,由样本建立分类规则库,通过计算分类数据规则条件部分与分类规则库中规则条件部分的模糊贴进度进行土地利用分类。结果表明:与传统的最大似然法分类方法相比,基于模糊规则的分类方法在高模糊性数据分类中显著提高了分类精度,在低模糊性数据分类中也能取得与最大似然法近似的结果。

In order to represent vague and imprecise value and improve the classification accuracy of remote sensing images,a fuzzy rule-based classification method was proposed. Firstly,by transforming the fuzzy membership function values into corresponding fuzzy subsets, fuzzy rule conditions were established. And then, the fuzzy rule database was derived from samples. Finally, based on the fuzzy nearness degrees of rule conditions were calculated from classified data and fuzzy rule database, the land use was classified. The ex perimental results show that the proposed method is able to significantly improve the classification accuracy than the maximum likelihood method while the data contains complex mixture of spatial information. Fur thermore, this method can get the approximate results as the maximum likelihood method while the data contains relatively homogeneous spatial information.