A new system for automatic classification of Landsat TM, based on Supervised ART-I and Supervised ART-II, has been develoepd. The sensitivity of the system has been tested for all domains of dynamic parameters: the vigilance parameter #rho# [0,1], and the dynamic learning parameter #beta# [0,1]. This has been achieved due to the simple architecture of Supervised ART-I and Supervised ART-II. The system has been applied for classifying a scene (256 x 240 pixels) of a Landsat TM image. The scene corresponds to the area around the Spanish city "Talavera de la Rina". Training the data with 9000 exemplars, using #rho#=0.98 and #beta#=0.50, leads to 85.82
展开▼
机译：已经开发了一种基于监督的ART-I和监督的ART-II的Landsat TM自动分类新系统。系统的灵敏度已针对动态参数的所有域进行了测试：警惕性参数＃rho＃[0,1]和动态学习参数＃beta＃[0,1]。这是由于受监管的ART-I和受监管的ART-II的简单体系结构而实现的。该系统已应用于对Landsat TM图像的场景（256 x 240像素）进行分类。该场景对应于西班牙城市“ Talavera de la Rina”周围的区域。使用＃rho＃= 0.98和＃beta＃= 0.50用9000个示例训练数据，得出85.82
展开▼
