摘要
以黑龙江省南瓮河湿地自然保护区内的SPOT5全色与多光谱遥感影像为研究对象,利用ENVI4.8、Arc-gis10.0遥感影像处理软件对4种常用融合方法进行质量评价。结果表明:Gram-Schmidt Spectral Sharpening变换融合方法效果最好,灰度均值和标准差最接近原多光谱影像,信息熵高于原多光谱,而且还保留了与SPOT5影像相同的4个波段,可以根据需要进行不同的波段选择;PC Spectral Sharpening变换后也保留4个波段,但这种融合方法色彩的饱和度较差,没有更好地增强部分地物;HSV变换和Brovey变换后图像只有3个波段,HSV变换后的色彩差异与源图像最大,不同地物间色彩差异不大,不利于地物的分类时的目视解译。
The SPOT-5 PAN and multi-spectral images of Nanwenghe Wetland National Nature Reserve in Hei- longjiang Province were selected as the original fusion data. Four fusion methods were quantitatively analyzed by ENVI4.8 ＆ Arcgisl0.0 image-processing software for remote-sensing. Result shows that：transform fusion method of Gram-Schmidt Spectral Sharpening is optimal. Gray mean ＆ Standard deviation are closed to the original multi- spectral image. Information entropy is higher than that of the original multi-spectral, retained four bands to different choices, the same as SPOT-5 images. PC Spectral Sharpening also retained four bands after transformation;this kind of fusion method whose color saturation is poor havn＇t optimally strengthened parts of ground objects. The image has only three bands after HSV ＆ Brovey transform;the color difference after HSV transform are the most signifi- cant;the color difference has little difference among different ground objects; there was no advantage to the visual in- terpretation of classified ground objects.
出处
《防护林科技》
2013年第2期30-34,共5页
Protection Forest Science and Technology
基金
国家公益性行业专项经费项目(201004074)
