基于像斑的多光谱影像跨尺度分类研究

基金项目: 国家973计划资助项目(2006CB701303,2004CB318206)；国家自然科学基金资助项目(40371079)

详细信息

Application of Decision Tree on Multispectral Images Based on Segment and Scale-Span Features

¹School of Remote Sensing and Information Engineering,Wuhan University,129 Luoyu Road,Wuhan 430079,China;²State Key Laboratory for Information Engineering in Surveying,Mapping and Remote Sensing,Wuhan University,129 Luoyu Road,Wuhan 430079,China

Funds: 国家973计划资助项目(2006CB701303,2004CB318206)；国家自然科学基金资助项目(40371079)

摘要: 提出了一种新的多尺度像斑模型,充分利用多尺度像斑模型所提供的尺度纵向信息,并结合决策树的分类方法来实现跨尺度分类,而不直接进行最佳尺度选择。实验结果证明,跨尺度方法较单一尺度分类能更准确地区分地物,从而提高分类精度。
- 多光谱 /
- 影像分割 /
- 像斑分析 /
- 决策树 /
- 多尺度
Abstract: Existing classification methods,based on the homogeneous-region,mostly involve the best segmentation criterion choice.Using the so-called best-scale to classify the multi-scale objects defined by human subjectivity,the paper doesn't think it is the best way for classification.So the paper proposes a new multi-scale homogeneous-region model,fully using the longitudinal information which the homogeneous-region model provides,and adopting the scale-span classification method based on decision tree to improve the accuracy,without selecting the best-scale data.The result shows this method can distinguish objects accurately and improve the precision than sole scale classification.
- multispectral image /
- segmentation /
- segment /
- decision tree /
- scale-span