北京航空航天大学学报 ›› 2010, Vol. 36 ›› Issue (9): 1103-1107.

• 论文 • 上一篇    下一篇

基于新息正交的工业CT图像自适应缺陷识别

左 凯, 孙同景, 李振华, 陶 亮   

  1. 山东大学 控制科学与工程学院, 济南 250061
  • 收稿日期:2010-03-12 出版日期:2010-09-30 发布日期:2010-10-08
  • 作者简介:左 凯(1972-),男,山东泰安人,博士生,zuokai@tom.com.
  • 基金资助:

    国防科工委基础科研基金资助项目(B142008.0209-08)

Adaptive industrial CT outlier detection and accommodation using innovation orthogonal

Zuo Kai, Sun Tongjing, Li Zhenhua, Tao Liang   

  1. School of Control Science and Engineering, Shandong University, Jinan 250061, China
  • Received:2010-03-12 Online:2010-09-30 Published:2010-10-08

摘要: 针对X射线投影图像的特点,研究基于Kalman滤波算法新息正交原理的缺陷识别问题.建立了投影图像的二维状态空间模型,定义了确定图像区域大小与灰度的动态评价函数和相应的阈值.在分析Kalman滤波算法新息正交性的基础上,得到了一种新的缺陷判别方法,提出了图像数据自适应补偿算法,提高了图像的识别效率和精度,并通过实验验证了方法的有效性.

Abstract: The problem of defect recognition in industrial CT(computerized tomography) projection image was discussed based on the innovation orthogonality principle. Projection image was modeling in a 2-D state space structure. By analyzing the innovation orthogonality of Kalman filtering algorithm, a dynamic evaluation function and the corresponding threshold on the determination of image region and grey scale were defined respectively. A new adaptive defect recognition method was obtained. Based on 2-D Kalman filter, the proposed method was implemented for defecting image recognition adaptively which can improve the efficiency and accuracy for CT image recognition. An experiment was given to illustrate the effectiveness of the proposed method.

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