基于深度学习的地物要素智能提取系统设计  

Design of an Intelligent Extraction System for Ground Feature Elements Based on Deep Learning

作　　者：胡腾飞 郑灿辉 袁红 王德斌 HU Tengfei;ZHENG Canhui;YUAN Hong;WANG Debin(Heilongjiang Third Surveying and Mapping Engineering Institute,Harbin 150025,China)

机构地区：[1]黑龙江第三测绘工程院,黑龙江哈尔滨150025

出　　处：《测绘与空间地理信息》2023年第S01期142-144,148,共4页Geomatics & Spatial Information Technology

摘　　要：随着人工智能在图像识别领域的应用不断深入,传统的地物分类正从人工采集向自动化/半自动化的模式发展。本文主要开展利用多源遥感影像数据基于深度学习算法实现植被、水体、道路、居民地等典型地物的智能化提取的研究,进行模块设计,为用户提供可操作、可视化的系统软件设计。本系统设计立足应用实际,遵循模块化、可拓展的设计原则,提供水系、道路、植被、居民地等典型地物智能提取解决方案,以减少人工投入,提高生产效率。With the continuous development of artificial intelligence in the field of image recognition,traditional feature classification is developing from manual collection to automated/semi automated mode.This article mainly conducts research on the intelligent extraction of typical features such as vegetation,water bodies,roads,residential areas,etc.based on deep learning algorithms using multi-source remote sensing image data,and conducts module design to provide users with operable and visual system software design.The design of this system is based on practical applications,following the principles of modularization and expandability.It provides intelli-gent extraction solutions for typical features such as water systems,roads,vegetation,residential areas,etc.,in order to reduce manual investment and improve production efficiency.

关 键 词：深度学习 地物要素 遥感影像 智能提取 

分 类 号：P209[天文地球—测绘科学与技术]