摘要
针对利用光谱混合分解提取不透水面特征通常受到端元类型和数量的限制,同时植被变化会影响估计精度的问题,该文提出了一种综合季相和植被变化信息的不透水面提取框架。基于混合像元部分分解算法——混合调谐匹配滤波(MTMF),设计了多季相组合MTMF(SCMTMF)特征和多季相叠加MTMF(SSMTMF)两种策略,构造了不透水面的多季相MTMF特征,将不透水特征与多季相植被指数结合利用支持向量机实现对不透水面的精确分类。结果表明,利用多季相特征得到的不透水面提取效果相较于单季相有较明显的改善,该文所提出的策略有利于提高不透水面的估计精度。
For the problems that features of impervious surfaces extracted by spectral mixture analysis(SMA) are usually limited by the determination of the endmembers, and the estimation accuracy can be affected by vegetation change, a framework of impervious surface extraction based on multi-season information was proposed in this paper. Based on mixture tuned matched filtering(MTMF) partial unmixing algorithm, multi-season MTMF features of the impervious surfaces were constructed according to two different strategies: seasons combined MTMF(SCMTMF) feature and seasons stacked MTMF(SSMTMF) feature. Then, a multi-season vegetation index was introduced to optimize the features to achieve accurate classification of impervious surfaces. Support vector machine was used to classify the impervious surface area(ISA) of the study areas based on multi-season MTMF features. Improvements were achieved using multi-season imagery, indicating that the proposed strategy could improve the estimation accuracy of impervious surface areas.
作者
陈姣
黄远程
李朋飞
CHEN Jiao;HUANG Yuancheng;LI Pengfei(College of Geomatics,Xi’an University of Science and Technology,Xi’an 710054,China)
出处
《测绘科学》
CSCD
北大核心
2021年第4期90-99,共10页
Science of Surveying and Mapping
基金
国家自然科学基金项目(41807063)。
