摘要
为了充分利用中等分辨率遥感影像的空间、纹理和光谱信息,解决影像中存在的大量混合像元问题,提出了基于地理分区的多端元光谱混合分析方法。实验结果表明,基于地理分区的多端元光谱混合分析方法消除了多端元光谱混合分析方法造成的像元过度分解现象,能够得到比较连续且和实际地物分布较为吻合的分析结果,其分解精度和效率优于多端元光谱混合分析方法。
In order to make full use of medium spatial resolution remote sensing images＇ space,texture and spectral information and solve the problems that mixed pixels widely existing in images,the method of multiple endmember spectral mixture analysis based on geographical division is proposed in this paper. The results show that multiple endmember spectral mixture analysis based on geographical division eliminates the phenomenon of unmixing the pixels excessively,achieves the analyzing results more continuous and more coincided with the actual distribution of surface features,and gets higher accuracy and decomposing efficiency than the multiple endmember spectral mixture analysis.
出处
《测绘地理信息》
2015年第4期63-66,共4页
Journal of Geomatics
基金
山西农业大学科技创新基金资助项目(20142-23)
关键词
地理分区
多端元光谱混合分析
TM遥感影像
影像分类
geographical division
multiple endmember spectral mixture analysis
TM remote sensing images
image classification
