近年来,在遥感图像计算机识别与分类方面已经作了相当大的努力。但是大部分的工作集中在逐点分类方法的研究。卫星遥感图像,由于大气传输及其各种环境因素的影响,许多地物在一定波谱段内具有相近的电磁辐射特性,致使不同类别物体在光谱空间相互交迭,因此,只利用单个像元光谱亮度值的分类方法,可能会造成较大的误分。 In recent years, considerable efforts have been made in computerized recognition and classification of remote sensing images. However, much of the work has focused on the study of point-by-point classification methods. Due to the influence of atmospheric transmission and various environmental factors, many objects have similar characteristics of electromagnetic radiation in certain spectral bands, causing different types of objects to overlap each other in the spectral space. Therefore, using only a single pixel spectrum The classification of brightness values, may cause greater misclassification.