论文部分内容阅读
受同物异谱和异物同谱现象的影响,对遥感影像进行分类时若仅利用光谱信息则分类精度的提高将会受到限制,而局部空间统计特征可以通过对地物空间聚集度的描述与分析在一定程度上减轻这种影响。本文研究了局部空间统计在不同指数(Moran’s I, Getis-Ord Gi, Geary’s C)、邻域规则和间隔距离下,对高空间分辨率的SPOT 5影像分类精度的影响规律。首先,对波段1进行局部空间统计分析,运算结果作为纹理波段添加到原始的光谱波段中;然后,综合利用光谱波段和纹理波段进行监督分类;最后,选取测试样本进行分类的精度评价,并比较分析不同条件下的分类精度,得到地物分类精度同参数之间的关系与规律。通过分析可以得出Getis-Ord Gi指数对于总体分类精度的提高最理想,总体分类精度从 87.74%提高到95.12%。
Affected by the phenomenon of the same matter and the foreign matter, the classification accuracy of the remote sensing image will be limited if only the spectral information is used. However, the statistical characteristics of the local space can be described by The analysis mitigates this effect to some extent. This paper studies the influence of local spatial statistics on the classification accuracy of SPOT 5 images with high spatial resolution under different indices (Moran’s I, Getis-Ord Gi, Geary’s C), neighborhood rules and separation distances. Firstly, the local spatial statistical analysis of band 1 is performed, and the result of the calculation is added as the texture band to the original spectral band. Then, the spectral band and the texture band are used for supervised classification. Finally, the test samples are selected to evaluate the accuracy of the classification and compared Analyze the classification accuracy under different conditions, and get the relationship and regularity between the classification accuracy and the parameters. The analysis shows that the Getis-Ord Gi index has the best overall classification accuracy, and the overall classification accuracy has increased from 87.74% to 95.12%.