基于LBV数据变换方法的海岸带多光谱影像分类方法研究

Research and Evaluation of Coastal Zone Multi-spectral Image Classification Based on LBV Transformation

在传统的海岸带多光谱影像分类方法的基础上,提出了运用LBV数据变换方法对海岸带多光谱影像进行地物分类识别的方法,通过图像数据变换能够分别提取出表示地物总辐射水平、可见光-近红外辐射平衡和表示波段辐射变化矢量(方向和速度)L、B、V的值。通过应用青岛海岸带TM多光谱影像,提取LBV变换图像并与原始假彩色合成图像比较,分别进行监督分类试验。运用LBV数据变换方法获得的图像信息量更丰富,地物类别更鲜明,有利于图像的目视解译,分类精度提高约3.12%。

Based on the traditional classification methods of coastal zone multi-spectral images,a new method using the LBV data transformation to process multi-spectral images of coastal zone before classified was proposed.LBV data transformation is a new method to process remote sensing images,by which,three basic remote-sensing characteristics of the general radiance level L,the visible-infrared radiation balance B,and the band radiance variation vector（direction and speed） V can be acquired respectively.Experiments for TM multi-spectral images cover Qingdao coastal zone was carried out,contrastively,the color composite of the digital L,B and V images transformed from original TM2,TM3,TM4 and TM5 images contains visually richer contents,more vivid ground features.The results of classification of the color composite using LBV transformation improved the classification accuracy about 3.12% than that of the original TM2,TM3,TM4 images,which showed that LBV transformation has good potential in coastal zone multi-spectral images classification.