作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2007, Vol. 33 ›› Issue (06): 198-200.

• 人工智能及识别技术 • 上一篇    下一篇

一种复杂场景下建筑目标识别方法

金泰松1,叶聪颖2,李翠华2,刘明业2   

  1. (1. 北京理工大学计算机科学工程系,北京 100081;2. 厦门大学信息科学与技术学院,厦门 361005)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-03-20 发布日期:2007-03-20

Approach to Building Recognition in Complex Scenes

JIN Taisong1, YE Congying2, LI Cuihua2, LIU Mingye2   

  1. (1. Department of Computer Science and Engineering, Beijing Institute of Technology, Beijing 100081; 2. School of Information Science and Technology, Xiamen University, Xiamen 361005 )
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-03-20 Published:2007-03-20

摘要: 提出了一种基于建筑目标的竖直线特征寻找图像中存在建筑目标区域的方法;考虑了目标特征的相互关系,给出了一种新的模板匹配算法。实验表明:利用该文提出的算法建立的识别系统与其它识别系统相比,大大减少了运算时间,有较好的抗噪声干扰和处理目标被遮挡问题的能力。

关键词: 竖直线, 模板匹配, 建筑提取, 建筑识别

Abstract: A new approach to building extraction based on grouping the feature of the vertical lines of building object is proposed. Using it, the regions where building may exist are marked. A new template matching algorithm, which accounts for the dependencies between features of object, is presented. Tests show that the recognition system can save a lot of runtime, and yield substantial improvement over others for cluttered scenes and processing the partially occluded objects.

Key words: Vertical line, Template matching, Building extraction, Building recognition