摘要
本文提出了一种基于GIS领域空间知识的影像纹理尺度提取方法:该方法分析了GIS领域知识辅助下人类对纹理特征的认知过程,对4个步骤的具体实现算法进行探讨;并选取海南省昌江县作为研究区域,实验证明了本方法的有效性,且与传统的枚举法相比,在性能相近的情况下其效率显著提高,更适合于大数据量遥感影像分类运算。
Texture is an important image special knowledge, whereas scales are the major factor to affect its validity. This study put forward an algorithm of GIS spatial knowledge-based multi-scale texture feature extraction. It analyzed the human cognition custom for texture feature under GIS knowledge auxiliary, and concluded that the texture scale parameter is strong correlated with feature geom- etry characteristic, and it could be determined based on stable features geometry characteristic from a historical database. Based on the above, this study first described the spatial geometry characteristic by ER algorithm, then followed extracting texture scale parameters by relevant data mining algorithm. Finally, the strong correlation charistiristics between texture scale factor and classified object shape were proved in experiments that the proposed algorithm could be used to quickly and adaptively choose the texture scale factors compa- ring with the traditional methods, to meet the request. Furthermore, the multi-resolution texture descriptors could improve the separa- bility among different land use categories, so that the total number of classification categories could be increased and more in line with the land use requirements.
出处
《测绘科学》
CSCD
北大核心
2013年第2期109-112,127,共5页
Science of Surveying and Mapping
基金
国家973基金支持项目(2006CB701303)
关键词
高分辨率遥感影像分类
多尺度纹理特征提取
GIS空间知识
high-resolution remote sensing images classification
multi-scale texture feature extraction
GIS spatial knowledge
