论文部分内容阅读
如果使用资源卫星图象对地物类型自动填图的方法要成为生产工具,就需要精确、迅速又经济地进行分类处理。通常所采用的最精确的分类方案,诸如最大似然判别法则。需要大量的计算。用传统的计算方法处理单帧资源卫星图象的费用很贵,阻碍了在生产的规模上对图象进行分类。设计了一个新的查表的方案,用最大似然高斯判别法则进行分类,减少了计算时间。通过对多光谱扫描器四个波段的光谱强度的高度相关性计算,把一个图象中特异的强度矢量的数目从1,600万千可能的矢量减少到为数千个。从而使得把特异的矢量和地物分类一起储存在计算机的磁心储存器中成为可能。这种查表寻找方案使资源卫星图象的信号分类比通常的方法至少快一个数量级而不损害其精确度,也不要求添加特定的计算机硬件。
If resource-satellite imagery-based methods of automatic mapping of terrain types are to be used as production tools, classification needs to be done accurately, quickly and economically. The most accurate classification schemes that are usually used, such as the Maximum Likelihood Decision Rule. Need a lot of calculation. The cost of processing satellite imagery of single-frame resources using conventional computational methods is prohibitively high, hindering the classification of images on a production scale. A new scheme of look-up table is designed, which is classified by the maximum likelihood Gaussian method to reduce the calculation time. By calculating the high correlation of the spectral intensities of the four bands of a multispectral scanner, the number of specific intensity vectors in an image is reduced from 16 million possible vectors to thousands. Making it possible to store specific vectors and features together in a magnetic core reservoir of a computer. This look-up table search scheme makes the classification of resource satellite imagery at least an order of magnitude faster than the usual methods without compromising its accuracy and without requiring the addition of specific computer hardware.