基于Landsat 8 OLI影像的大兴安岭西麓春小麦识别方法的比较研究

Study on comparison of identification methods for spring wheat at Great Khingan western slope based on Landsat 8 OLI

采用Landsat 8 OLI影像数据和野外调查数据,结合物候学、目视解译、非监督分类ISODATA法和监督分类最大似然法,以及面向对象分类法,对大兴安岭西麓农场的春小麦识别进行了初步探索。结果表明:最大似然法、ISODATA法和面向对象分类法的分类总体精度分别为53.33%、54.65%和71.26%;三种方法的春小麦用户精度分别为60.00%、61.29%和83.33%。在大兴安岭西麓苗期的春小麦识别中,面向对象分类法优于ISODATA法和最大似然法,为该区域春小麦识别提供思路。

Using Landsat 80LI image and field survey data, combined with visual interpretation, ISODATA classification of unsupervised classification, maximum likelihood classification of supervised classification and object-oriented image classification, a preliminary study was conducted on spring wheat identification at farmland in Great Khingan western slope. The results showed the whole accuracy of classification accuracy in maximum likelihood, ISODATA and object-oriented image classification were 53.33% ,54.65%;~ 71.26% respectively; user accuracy of spring wheat in three methods were 60.00%, 61.29% and 83.33% respectively. The object-oriented image classification was better than ISODATA and maximum likelihood classification, and it would provide ideas for spring wheat identification in these areas.