基于小波分量特征值匹配的高光谱影像分类

基金项目: 国家973计划资助项目(2003CB415205);国家自然科学基金资助项目(40471088405230051)

详细信息

Hyperspectral Image Classification Based on Features of Wavelet Vectors

¹ State Key Laboratory of Information Engineering in Surveying,Mapping and Remote Sensing,Wuhan University,129 Luoyu Road,Wuhan 430079,China;² Chengdu Institute of Survey and Investigation,70 Yihuan Road,Chengdu 610081,China

Funds: 国家973计划资助项目(2003CB415205);国家自然科学基金资助项目(40471088405230051)

摘要: 提出了一种基于小波分量特征值的高光谱影像分类算法。针对每个像素构建一个能反映该分量特征的函数,得到其特征值。再利用这些特征值与参考光谱的特征值进行匹配,从而对整幅影像实现分类。实验证明,该方法比传统的光谱角制图法和交叉相关系数法的分类精度有较大提高。
- 特征提取 /
- 图像分类 /
- 光谱匹配 /
- 小波变换
Abstract: A novel approach for mapping hyperspectral data is presented by extracting the feature of wavelet vectors.The expected decomposed level-L is decided for a given hyperspectral imagery and then a function is constructed to extract the feature of each vector for every pixel.Classified image is obtained by comparing the feature between the pixel vector and the reference spectra.Comparisons with spectral angle mapping(SAM) and cross correlation spectral match(CCSM) are also done.
- feature extraction /
- image classification /
- spectral matching /
- wavelet transform