西安电子科技大学学报

• 研究论文 • 上一篇    下一篇

一种自动的高分辨率遥感影像道路提取方法

刘如意1;宋建锋1;权义宁1;许鹏飞2;雪晴1;杨云3;苗启广1   

  1. (1. 西安电子科技大学 计算机学院,陕西 西安 710071;
    2. 西北大学 信息科学与技术学院,陕西 西安 710127;
    3. 西安测绘研究所 地理信息工程国家重点实验室,陕西 西安 710054)
  • 收稿日期:2016-01-20 出版日期:2017-02-20 发布日期:2017-04-01
  • 通讯作者: 苗启广(1972-) ,男,教授,E-mail:qgmiao@126.com
  • 作者简介:刘如意(1989-) ,女,西安电子科技大学博士研究生,E-mail:ruyi198901210121@126.com
  • 基金资助:

    国家自然科学基金资助项目(61472302,61272280,U1404620,41271447);教育部新世纪优秀人才支持计划资助项目(NCET-12-0919);中央高校基本科研业务费专项资金资助项目(K5051203020,JB150313,K5051303018, BDY081422);陕西省自然科学基金资助项目(2014JM8310, 2010JM8027);西安市科技局资助项目(CXY1441(1));地理信息工程国家重点实验室开放研究基金资助项目(SKLGIE2014-M-4-4)

Automatic road extraction method for high-resolution remote sensing images

LIU Ruyi1;SONG Jianfeng1;QUAN Yining1;XU Pengfei2;XUE Qing1;YANG Yun3;MIAO Qiguang1   

  1. (1. School of Computer Science and Technology, Xidian Univ., Xi'an 710071, China;
    2. School of Information Science and Technology, Northwest Univ., Xi'an 710127, China;
    3. State Key Lab. of Geo-information Engineering, Xi'an Research Institute of Surveying and Mapping, Xi'an 710054, China)
  • Received:2016-01-20 Online:2017-02-20 Published:2017-04-01
  • Contact: MIAO Qiguang

摘要:

从高分辨率遥感影像中提取道路有着非常重要的意义,但是受到遥感影像噪声、复杂的自然场景和已有算法的局限性的影响,道路提取有待于进一步研究.近些年来水平集方法被用于提取道路,但是初始水平演化曲线的确定却是一个大的难点.笔者提出一种自动的水平集分割方法,并将其用于道路检测中.首先,将卷积神经网络用于道路的粗分类.然后,利用形状特征和孔洞填充方法得到比较准确的道路区域.在此基础上,利用张量投票来提取道路的交叉口,并将其轮廓作为水平集演化的初始曲线进行水平集分割.最后,结合卷积神经网络分类和水平集分割的优势,得到比较完整的道路区域,并保持了道路的边缘.实验结果表明,该方法能自动地提取准确完整的道路区域.

关键词: 卷积神经网络, 形状特征分析, 张量投票, 水平集分割, 信息融合

Abstract:

Road extraction from high-resolution satellite images is very important. Due to image noise, the natural scene complexity, and the extraction algorithms limitations, it still needs to be further researched. In recent years, level set evolution has been used to extract the road, but it is difficult to automatically generate initial level curves for the level set evolution (LSE). In this paper, we propose an automatic approach to the generation of initial level curves and use it to extract the road. Firstly, the convolutional neural network(CNN) is used to classify the road or nonroad, then shape features are adopted to filter nonlinear features to get the accurate road region. And on this basis, we exploit tensor voting to detect the road junctions and utilize them as initial level curves; finally we fuse the results obtained by the CNN and LSE. Experiments show that this algorithm can get an accurate and complete road.

Key words: convolutional neural network(CNN), shape feature, tensor voting, level set, information fusion