基于无人机与卫星高分辨率光学影像的公路地震灾害检测关键技术研究

The unmanned aerial vehicle (UAV) remote sensing system has the characteristics of low-cost, flexibility, avoiding cloud interference, can obtain high-quality and high-resolution images after the earthquake, which has become an effective means to obtain earthquake disaster information. This project intends to use the high-resolution optical image of the UAV after the earthquake and the high-resolution optical satellite image before the earthquake to achieve rapid detection of highway earthquake disasters. This research focuses on solving the three typical problems: low efficiency of UAV image mosaic, difficulty in registration for heterogeneous images between UAV and satellite, and low efficiency of earthquake disaster information detection. The study includes three parts: (1) Research on high-resolution optical image mosaic of UAV based on raster computing and image segmentation to achieve high-quality rapid mosaic of high-resolution UAV images; (2) Research on the multi-featured and progressive precise registration for high resolution optical image from UAV and satellite which based on change detection，to realize high-accuracy image registration for heterogeneous high-resolution images, and provide a solid support for highway earthquake disaster detection. (3) Research on the rapid detection method of object-oriented highway earthquake disaster based on scale set theory, to improve the accuracy and efficiency of highway earthquake disaster detection. Through the above research, it will break through the key scientific problems in the rapid detection of highway earthquake disasters and provide traffic emergency protection for earthquake relief, which has great theoretical significance and application value.

无人机遥感系统成本低、机动灵活、能避开云层干扰，获取震后高质量高分辨率影像，已成为获取地震灾害信息的有效手段。本项目拟利用震后无人机高分辨率光学影像，并结合震前高分辨率光学卫星影像实现对公路地震灾害的快速检测。重点解决无人机影像镶嵌效率低，无人机与卫星异源影像配准难度大、地震灾害信息检测效率低三个典型问题。具体研究内容包括：（1）研究基于栅格计算及影像分割的无人机高分辨率光学影像镶嵌，实现高分辨率无人机影像的高质量快速镶嵌；（2）研究融合变化检测的无人机与卫星光学影像多特征渐进式精配准，实现异源高分辨影像的精配准，为公路地震灾害检测奠定基础。（3）研究基于尺度集理论的面向对象公路地震灾害快速检测方法，提高公路地震灾害检测的精度和效率。通过上述研究，突破公路地震灾害快速检测中关键科学问题，为抗震救灾提供交通应急保障，有重要理论意义和应用价值。

本项目利用震后无人机高分辨率光学影像，并结合震前高分辨率光学卫星影像实现对公路地震灾害的快速检测。重点解决无人机影像镶嵌效率低，无人机与卫星异源影像配准难度大、地震灾害信息检测效率低三个典型问题。.通过三年的研究工作，本项目取得了如下主要研究成果：（1）提出一种基于栅格计算的初试接缝线网络生产方法，分别提出基于影像分割的对象级与像素级相结合的镶嵌线优化和基于D-LinkNet神经网络道路概率图的镶嵌线优化方法，实现高分辨率无人机影像的高质量快速镶嵌；（2）将变化检测引入到精配准中，提出一种融合变化检测的无人机与卫星光学影像多特征渐进式精配准方法，实现异源高分辨影像的精配准，为公路地震灾害检测奠定基础。（3）将尺度集理论引入地震灾害检测中，并结合已有交通基础数据，提出一种基于尺度集理论的面向对象公路地震灾害快速检测方法，提高公路地震灾害检测的精度和效率。通过上述研究，突破公路地震灾害快速检测中关键科学问题，为抗震救灾提供交通应急保障，有重要理论意义和应用价值。.项目执行期间内，项目负责人以项目研究最新成果，共发表学术论文4篇，其中SCI论文2篇，核心期刊论文2篇；培养相关人才4名；申请国家发明专利4项，其中已授权1项；基于本项目研究成果发表的论文1篇入选2021年度交通运输重大科技创新成果库入库成果（科技论文类）。