摘要
利用模式识别中特征提取和特征选择的相关理论对遥感影像的纹理特征进行遴选和变换处理,得到描述纹理的二次特征。实验证明这些新特征能够提高影像分类精度和分类效率。
In this paper, some problems in image texture are studied making use of the theories of feature extraction and selection.15 texture features are gained through gray-level histogram, co-occurrence matrix and run-length matrix.2～3 new features have been chosen out of these 15 texture features and have done a better work on image classification.
出处
《测绘科学》
CSCD
北大核心
2008年第1期176-178,199,共4页
Science of Surveying and Mapping
关键词
特征提取
特征选择
影像分类
纹理特征
feature extraction
feature selection
image classification
image texture
