遥感数据专题分类不确定性评价研究:进展、问题与展望

ASSESSMENT ON UNCERTAINTY IN REMOTELY SENSED DATA CLASSIFICATION: PROGRESSES, PROBLEMS AND PROSPECTS

从遥感数据中提取专题类别信息是当前遥感数据最主要的应用领域之一。由于遥感分类专题信息广泛应用于各种领域,其数据质量受到越来越多的关注。不确定性是评价分类专题类别数据质量最主要的方面。回顾了遥感数据专题分类不确定性评价方法的历史,总结了当前各种评价方法及其指标体系,将这些方法归结为基于误差矩阵的方法、模糊评价方法、像元尺度上的不确定性评价方法和其它方法四大类。对每一类不确定性评价方法及其指标体系的优点和缺点进行了分析和总结,指出从理论方法研究方面,需要优先发展独立于分类方法的像元尺度上的遥感分类不确定性评价模型与指标体系,以及统一的遥感数据分类不确定性评价模型体系研究;在应用研究方面,需要加强优化空间采样设计和不确定性评价过程标准化研究。

Thematic mapping is one of the major application fields of remote sensing techniques. Due to the wide use of the thematic data derived from remotely sensed data by either visual interpretation or automatic classification, data quality of the thematic data was concerned about. Uncertainty in the classified remotely sensed data is one of the most important elements of the data quality. In this paper, the development of the methods for assessment of uncertainty in remotely sensed data classification was overviewed, existing methods and the uncertainty measurements were reviewed and categorized into error matrix based methods, fuzzy set based methods, methods of assessment at pixel scale and other methods. Advantages and the problems remained in every kind of method were analyzed. Future research was suggested that, from the perspective of theoretic study, the uncertainty assessment method at pixel scale and the uncertainty measurements independent on the classifier should be developed; from the perspective of application, attentions should be paid on the optimal spatial sampling and the standardization of uncertainty assessment procedures.