证据推理遥感分类背景下遥感影像证据支持度生成方法的对比

摘要

选择黑龙江省洪河自然保护区的ETM+遥感影像,对比分析了基于最小距离法(MD)和频率分布法(FD)两种不同证据支持度生成方法的证据推理遥感分类精度.结果表明,在使用前3个或更少的波段情况下,MD方法证据支持度质量比MD方法好,而在前4个或更多的波段情况下,MD方法生成的证据支持质量更高.证据支持度的合理与否直接影响证据推理的分类精度和效果,而MD方法在多波段情况下在证据支持度生成方面更具优势.%A scene of ETM+ imagery covered Honghe National Nature Reserve was chosen to compare the ability of two algorithms, the minimum distance (MD) algorithm and the frequency distribution (FD) algorithm, in generation EM for supervised evidential classification. Results show that the FD algorithm is superior to the MD algorithm when using the first three or fewer bands of ETM+ imagery, while the MD algorithm is better when using the first four or more bands for classification. It indicates that the EM generation algorithm directly affects the accuracy of the supervised evidential classification, and the MD algorithm has advantages when using multi-bands of ETM+ imagery for EM generation.