基于深度学习的新增违法建设监测框架研究与应用  

Research and application of deep learning-based framework for monitoring and detecting new illegal construction

作　　者：康停军 陈汭新 孙颖[2] 夏义雄 王彬 KANG Tingjun;CHEN Ruixin;SUN Ying;XIA Yixiong;WANG Bin(Foshan Surveying Mapping and Geoinformation Research Institute,Foshan 528000,China;School of Geography and Planning,Sun Yat-Sen University,Guangzhou 510275,China)

机构地区：[1]佛山市测绘地理信息研究院,广东佛山528000 [2]中山大学地理科学与规划学院,广东广州510275

出　　处：《测绘通报》2023年第11期116-121,共6页Bulletin of Surveying and Mapping

基　　金：国家自然科学基金面上项目(42071441);广东省基础与应用基础区域联合基金-青年基金项目(2020A1515110441)。

摘　　要：科学化、智慧化严控新增违法建设是城市高质量发展的必然要求。针对新增违法建设类型多样、隐蔽性强、管理交叉、多头执法等难点,本文基于高分辨率遥感影像,利用耦合FPN与Mask-RNN多任务的建筑物边界提取算法,构建了“建设行为监测—多源数据融合分析—协同分派治理”的新增违法建设全链条管理框架,提升了高空视角下违法建设的精准发现、动态管理能力,为助力城市精细化管理提供了技术支撑。Scientific and intelligent control of new illegal constructions is an inevitable requirement for high-quality urban development.Aiming at the characteristics of new illegal constructions with diverse types,strong concealment,administrative intersection and difficult disposal,this paper adopts a building boundary extraction algorithm based on high-resolution remote sensing images and management data of functional departments,coupled with FPN and Mask-RCNN multi-task fusion.It constructs a whole chain management framework of new illegal constructions from“Construction Behavior Monitoring-Multi-source Data Fusion Analysis-Collaborative Assignment Governance”,which provides technical support for dynamic monitoring and precise management under the high-altitude perspective of the city.

关 键 词：新增违法建设 深度学习 高分辨率影像 全链条管理框架 

分 类 号：P237[天文地球—摄影测量与遥感]