基于多源遥感影像的智能解译样本库建设

Construction of intelligent interpretation sample database based on multi sourced remote sensing images

针对因深度学习样本数量和类型有限而导致的遥感影像的自动解译能力不足等问题,本文在分析现有样本集现状及问题的基础上,结合甘肃省地形地貌条件和已有数据,研究和探讨了甘肃省遥感影像解译样本库构建的总体思路和基本策略,建立了拥有百万级样本的遥感影像智能解译样本库。

For the problems such as poor abilities of automatic remote sensing images interpretation caused by limited number and types of deep learning samples,general ideas and basic strategies for the construction of remote sensing image interpretation sample database in Gansu Province were studied and discussed in this paper based on the analysis of the current situation and problems of the existing sample set,combined with the topographic and geomorphic conditions of Gansu Province and the existing data.A sample database of intelligent interpretation of remote sensing images with millions of samples is established.