面向矿山监管的无人机视频关键帧影像动态提取方法  

Dynamic Extraction Method of Key Frame Image of UAV Video for Mine Supervision

作　　者：张航 卢小平[1] 张晓强 路泽忠 ZHANG Hang;LU Xiaoping;ZHANG Xiaoqiang;LU Zezhong(Key Laboratory of Mine Spatial Information and Technology of NASMG,Jiaozuo,Henan 454003,China;Reconnaissance,Planning,Design and Research Institute,YRCC,Zhengzhou 450003,China)

机构地区：[1]河南理工大学矿山空间信息技术国家地理信息局重点实验室,河南焦作454003 [2]黄河勘测规划设计研究院有限公司,郑州450003

出　　处：《遥感信息》2020年第1期112-116,共5页Remote Sensing Information

基　　金：2016年国家重点研发计划(2016YFC0803103);河南省高校创新团队支持计划(14IRTSTHN026)。

摘　　要：为满足对露天煤矿违法开采实时监测需求,提出了一种基于无人机瞬时航飞速度的视频关键帧影像动态提取方法。首先对无人机采集的视频影像进行预处理,提取视频影像中无人机瞬时飞行速度、飞行航高等信息;然后基于提取的相关信息加权化处理并结合相机参数进行计算,得出抽取关键帧所需的动态间隔,再对关键帧时间间隔进行加权和阈值约束处理,提取出关键帧的动态时差。实例验证表明,采用该方法提取的关键帧影像进行矿区三维重建,在保证三维场景的同时有效提高了数据处理速度,且比等时差法用时缩短了近5倍,显著提高了三维重建的效率。In order to meet the requirement of real-time monitoring for illegal mining in open pit coal mines,a dynamic extraction method of video key-frame image based on the instantaneous flight speed of UAV is proposed in this paper.The method firstly preprocesses the video images collected by UAV to extract the instantaneous flight speed and flight height of UAV in the video images,then it calculates the dynamic interval needed to extract the key frame by weighted processing based on the extracted information and combining with camera parameters.Finally,the key frame time interval is weighted and the threshold constraint is processed to extract the dynamic time difference of the key frame.The key frame image extracted by this method is used for 3D reconstruction of mining area,which not only guarantees the 3D scene but also improves the data processing speed effectively,and shortens the time by nearly five times compared with the equal-time difference method,obviously improves the efficiency of 3D reconstruction.

关 键 词：无人机 视频流 关键帧 三维重建 地理编码 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]