摘要
采用小波变换、小波和PCA相结合、小波和HIS相结合等遥感影像融合方法对Landsat7 ETM+多光谱与全色影像进行融合和土地覆盖分类研究,并结合影像的光谱统计参数和融合影像分类精度对这些方法进行评价,结果发现这3种融合方法均能不同程度地提高影像的分类精度。其中,第3种方法所得融合影像与原多光谱影像的相关系数最大,均方差、相关系数和信息熵最大,影像所含信息量最多,光谱特性保持较好;清晰度较高,在空间细节信息的表现能力上较优,所以融合影像的分类精度最高。因此,小波和HIS相结合的融合方法更适合ETM+融合影像的土地覆盖分类研究。
The remote sensing image fusion methods of wavelet transform, combination of wavelet and PCA, and wavelet and HIS ete were used to fuse Landsat7 ETM ＋ multi-spectral and panchromatic images and study the land cover classification. And these methods were evaluated by combining the spectral static parameter of images with the classification accuracy of fused images. It was found in the results that all these 3 fusion methods could enhance the classification accuracy of images to different degrees, in which, the correlation coefficient between the fused images obtained by the 3rd method and original multi-spectral images was biggest, with the mean square deviation and information entropy being biggest, the information contained in images being most and spectrum characters being kept better. The fused images of the 3rd method had higher definition and superior capability on showing detail spatial information, so the classification accuracy of fused images was highest. Therefore, the fusion method of combining wavelet with HIS was more suitable for studying land cover classification of ETM ＋ fused images.
出处
《安徽农业科学》
CAS
北大核心
2008年第1期62-65,共4页
Journal of Anhui Agricultural Sciences
基金
高等学校博士学科点专项科研基金(20060290511)
