近红外光谱主成分分析与模糊聚类的典型地面目标物识别

发布时间：2018-06-21 14:30

本文选题：近红外光谱 + 主成分分析　；参考：《光谱学与光谱分析》2017年11期

【摘要】：近红外光谱技术在遥感监测领域中应用广泛,针对典型地面目标物遥感监测识别需要,提出了光谱主成分分析(PCA)与模糊聚类结合的分类识别方法,提高了识别算法效率及准确性。以四类典型地面目标物作为研究对象,分别测量其在1 100~2 500nm范围内漫反射光谱,首先对漫反射光谱进行主成分分析,得到代表光谱特征的主成分分量,然后将其作为模糊聚类分析模型输入参数,计算样品主成分集合之间贴合度,最后利用择近原则对样品进行匹配分类。结果表明,主成分分析可以有效提取光谱特征并且降低数据维度,结合基于择近原则的模糊分类方法,可有效提高算法准确性与效率,为遥感光谱在地面目标物识别应用提供了有益的参考。
[Abstract]:Near-infrared spectroscopy (NIR) is widely used in the field of remote sensing monitoring. Aiming at the need of remote sensing monitoring and recognition of typical ground objects, a classification and recognition method combining spectral principal component analysis (PCA) and fuzzy clustering is proposed. The efficiency and accuracy of the recognition algorithm are improved. The diffuse reflectance spectra of four typical ground objects were measured in the range of 1 100 ~ 2 500nm. The principal components representing the spectral characteristics were obtained by principal component analysis (PCA). Then it is used as the input parameter of the fuzzy cluster analysis model to calculate the sticker degree between the principal components of the samples. Finally, the matching classification of samples is carried out by using the principle of proximity selection. The results show that principal component analysis can effectively extract spectral features and reduce the data dimension. Combining with the fuzzy classification method based on the principle of selection and proximity, the accuracy and efficiency of the algorithm can be improved effectively. It provides a useful reference for the application of remote sensing spectrum in target recognition on the ground.
【作者单位】：天津大学精密测试技术及仪器国家重点实验室;天津大学精密仪器与光电子工程学院;
【基金】：国家(863)高技术研究发展计划(2012AA022602) 国家自然科学基金项目(81401454,81471698) 中国博士后科学基金项目(2013M541174) 中国博士后国际交流计划(20140066) 天津市应用基础与前沿技术研究计划(16JCZDJC31200)资助
【分类号】：O657.3;P237

【相似文献】