遥感影像分类结果的不确定性研究  

Research on Classification Uncertainties of Remote Sensing Image

作　　者：黄恩兴[1] 

机构地区：[1]天津市勘察院,天津300191

出　　处：《中国农学通报》2010年第5期322-325,共4页Chinese Agricultural Science Bulletin

摘　　要：遥感影像是获取地学信息的主要手段之一,遥感影像分类结果具有不确定性,数据获取、数据处理、分类算法、分类后处理及精度评价等均可能导致分类结果的不确定。此文主要从遥感数据质量的不确定性、分类算法的不确定性、分类评价方法的不确定性及分类后处理的不确定性4个方面讨论了遥感影像分类结果的不确定性。通常情况下,上述4个方面对分类结果的不确定性影响较大,需要对其深入了解才能提高分类结果的稳定性和分类精度。Various geographical information can be extracted based on remote sensing images, and the technique has become a main method in the analysis of geographical data. Normally, the classification results of remote sensing images have uncertainty and the uncertainty results from different aspects, such as data sources, data processing, classification algorithm, post-classification processing, and accuracy evaluation and so on. The classification uncertainties of remote sensing image were discussed in this paper, basically, including the uncertainty of remote sensing data quality, the uncertainty of classification algorithm, the uncertainty of classification assessment method, and the uncertainty of post-classification processing. Normally, the above four aspects have a serious influence on the classification uncertainty of remote sensing images, and only after they are fully realized, the accuracy and stability of classification can be improved.

关 键 词：遥感影像 分类 不确定性 分类后处理 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置] TP237.4[自动化与计算机技术—控制科学与工程]