北京航空航天大学学报 ›› 2015, Vol. 41 ›› Issue (3): 383-390.doi: 10.13700/j.bh.1001-5965.2014.0177

• 论文 • 上一篇    下一篇

基于MSER的无人机图像建筑区域提取

丁文锐1, 康传波2, 李红光1, 刘硕2   

  1. 1. 北京航空航天大学 无人驾驶飞行器设计研究所, 北京 100191;
    2. 北京航空航天大学 电子信息工程学院, 北京 100191
  • 收稿日期:2014-04-03 出版日期:2015-03-20 发布日期:2015-04-02
  • 通讯作者: 丁文锐(1971—),女(满族),辽宁鞍山人,研究员,ding@buaa.edu.cn,主要研究方向为遥感图像处理等. E-mail:ding@buaa.edu.cn
  • 基金资助:
    新世纪优秀人才支持计划资助项目; 总装预研基金资助项目(9140A25031112HK01303)

Building areas extraction basing on MSER in unmanned aerial vehicle images

DING Wenrui1, KANG Chuanbo2, LI Hongguang1, LIU Shuo2   

  1. 1. Research Institute of Unmanned Aerial Vehicle, Beijing University of Aeronautics and Astronautics, Beijing 100191, China;
    2. School of Electronic and Information Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100191, China
  • Received:2014-04-03 Online:2015-03-20 Published:2015-04-02

摘要: 对建筑区域自动检测与提取是无人机(UAV,Unmanned Aerial Vehicle)图像处理的一项重要功能.在分析无人机成像特点和最大稳定极值区域(MSER,Maximum Stable Extremal Regions)算法对无人机侦察图像建筑区域检测的适用性基础上,提出了一种基于MSER的无人机侦察图像建筑区域提取算法.算法包含5步:无人机图像预处理,运用MSER算法分析计算图像稳定区域,通过计算稳定区域密度筛选建筑区域,进一步利用自适应K均值聚类算法对建筑区进行划分,最后采用Graham算法生成建筑区的边界从而实现了建筑区的自动提取.选取无人机实飞图像数据进行实验统计,本算法提取精度为92.25%;同时与基于Gabor变换的纹理特征、SIFT特征点的提取算法相比,建筑区域提取时间缩短,满足无人机实时应用需求.

关键词: 建筑区域提取, 无人机图像预处理, 最大稳定极值区域, 自适应K均值聚类, Graham算法

Abstract: Automatic detection and extraction of the building area is an important aspect of unmanned aerial vehicle (UAV) image processing. Based on the detailed analysis of UAV imaging characteristics and the maximum stable extremal regions (MSER) algorithm, a building area extraction algorithm of UAV image is proposed. The algorithm consists of five steps: firstly, the pretreatment of UAV image; secondly, analysis and calculation of image stable regions using MSER; thirdly, screening the building area by calculating the density of stable regions; then, using adaptive K-means clustering algorithm to divide the building area; ultimately, boundaries of the building area were generated using Graham algorithm in order to achieve automatic extraction of building area. Using the UAV real flying image data to do the experiment statistics, the conclusion includes: Firstly, the extraction accuracy of this algorithm reaches 92.25%; secondly, when compared with other building area extraction algorithm which based on Gabor transform or SIFT, the extraction time of building area is shortened and meets the needs of UAV real-time applications.

Key words: building area extraction, pretreatment of UAV images, maximum stable extremal regions, adaptive K-means clustering, Graham algorithm

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发