计算机与现代化

• 图像处理 • 上一篇    下一篇

 民用航空飞机注册号自动识别

  

  1. 1.四川大学计算机学院,四川成都610065; 2.四川大学视觉合成图形图像技术国防重点学科实验室,四川成都610065
  • 收稿日期:2015-01-12 出版日期:2015-04-27 发布日期:2015-04-29
  • 作者简介:杨清帅(1991-),男,河南许昌人,四川大学计算机学院硕士研究生,研究方向:图像处理,计算机视觉; 杨红雨(1967-),女,四川成都人,四川大学视觉合成图形图像技术国防重 点学科实验室博士生导师,研究方向:计算机图像处理和图形学,计算机仿真,实时软件工程; 兰时勇(1974-),男,四川成都人,博士,研究方向:图形图像,计算机仿真; 程鹏(1985-),男,四川成 都人,博士,研究方向:图像融合,计算机视觉。
  • 基金资助:
     国家“863”计划项目(2013AA013802); 国家“863”计划项目(2012AA011804)

Registration Number Recognition of Civil Aviation Aircraft

  1. 1. College of Computer Science, Sichuan University, Chengdu 610065, China;

     2. State Key laboratory of Fundamental Science on Synthetic Vision, Sichuan University, Chengdu 610065, China
  • Received:2015-01-12 Online:2015-04-27 Published:2015-04-29

摘要:

为了提高机场自动化水平,提出一种简单、有效的飞机注册号自动识别算法。对于机场摄像机获取的彩色跑道图像,首先采用Canny操作和轮廓检测提取飞机图像,为了便于计算机处理将
其转换为灰度图像;然后使用Hough转换和SWT(Stroke Width Transform)文本定位法定位飞机尾部的注册号;最后应用阈值法分割注册号,使用基于孔洞的字符模板匹配法识别注册号字符。实验结果表
明,该算法具有良好的飞机注册号识别效果,较强的鲁棒性,具有广泛的应用前景。

关键词:  , 飞机注册号, 文本定位, 图像分割, 字符识别

Abstract:

In order to improve the automation level of airport, a novel algorithm is proposed for civil aviation aircraft registration number recognition. For the color
image of runway taken by cameras mounted in the aircraft, firstly the image of aircraft is extracted by Canny operation and contours detection, and then the color image is
converted into gray image for computer processing. Secondly, use the probabilistic Hough transform and Stroke Width Transform to position registration number. Finally, segment
registration number by the method of threshold and apply hole-based template matching method to recognize the characters of the registration number. Experimental results show
that the algorithm is feasible, robust and applicable.

Key words:  aircraft registration number, text positioning, image segmentation, character recognition