面向对象高分辨率影像信息提取中的尺度效应及最优尺度研究

Scale effect and optimal scale in object-oriented information extraction of high spatial resolution remote sensing image

本文从面向对象的遥感信息提取中的尺度效应研究入手,对影像对象的分形维数、紧凑度、面积、均值、标准差和与邻域均值差分等特征进行了实验。在此基础上,根据"类内同质性大,类间异质性大"的最佳分类原则,提出了面向对象的RMAS方法,该方法的思想是,当对象RMAS值最大时,对象内部的异质性最小,对象外部的异质性最大,此时的分割尺度为类别提取的最优分割尺度。根据最优尺度下信息提取精度最高的原理,实验验证了该方法的可行性,且能获得较好的分类结果。

From scale effect problem in the remote sensing information extraction, the targets＇ fractal dimensions, compact ratios areas, mean value, standard deviation and mean differential to neighbors of image objects were experimental researched in the paper. It found that these index values of all targets would fluctuate with scales, and different targets in the images have different feature Yalues and scales. It is necessary to extract the region of interest in optimal scale images. In view of this, RMAS method was developed, according to the best classification principle as ＂homogeneity in class, heterogeneity between classes＂ . The thought of this method was that the heterogeneity in class is the minimum and is the maximum between classes when RMAS is the maximum, so the segmentation scale is optimal. According to the principle of the highest information extraction accuracy based on the optimal scale, the experiment verified the feasibility of this method and the classification result was better.