中国组织工程研究

中国组织工程研究 ›› 2012, Vol. 16 ›› Issue (17): 3160-3163.doi: 10.3969/j.issn.1673-8225.2012.17.028

• 骨与关节图像与影像 bone and joint imaging • 上一篇    下一篇

基于图像分割的Tamura纹理特征算法的研究与实现***☆

吕晓琪，郭金鸽，赵宇红，任晓颖   

  1. 内蒙古科技大学信息工程学院，内蒙古自治区包头市  014010
  • 收稿日期:2011-07-11 修回日期:2011-12-19 出版日期:2012-04-22 发布日期:2012-04-22
  • 通讯作者: 郭金鸽，硕士，内蒙古科技大学信息工程学院，内蒙古自治区包头市 014010 yewen37@163.com
  • 作者简介:吕晓琪☆，男，1963年生，汉族，北京科技大学在读博士，教授，博士生导师，主要从事医学图像处理的研究。lxiaoqi@imust.cn
  • 基金资助:

    国家自然科学基金项目(61179019)，课题名称：基于物理模型的非刚性医学图像配准算法；内蒙古自然基金重点项目(2010Zd26)，课题名称：IHE框架下放射科数字化解决方案研究与实现；内蒙古科技大学创新基金项目(2010NC037)，课题名称：基于ITK-VTK的医学图像分割手法相关技术研究。

Research and realization of Tamura texture feature extraction method based on image segmentation 

Lü Xiao-qi, Guo Jin-ge, Zhao Yu-hong, Ren Xiao-ying   

  1. School of Information Engineering, Inner Mongolia University of Science and Technology, Baotou  014010, Inner Mongolia Autonomous Region, China
  • Received:2011-07-11 Revised:2011-12-19 Online:2012-04-22 Published:2012-04-22
  • Contact: Guo Jin-ge, Master, School of Information Engineering, Inner Mongolia University of Science and Technology, Baotou 014010, Inner Mongolia Autonomous Region, China yewen37@163.com
  • About author:Lü Xiao-qi☆, Studying for doctorate, Professor, Doctoral supervisor, School of Information Engineering, Inner Mongolia University of Science and Technology, Baotou 014010, Inner Mongolia Autonomous Region, China lxiaoqi@imust.cn
  • Supported by:

    National Natural Science Foundation of China, No.61179019*; Natural Science Foundation of Inner Mongolia Autonomous Region, No. 2010Zd26*; Innovation Fund of Inner Mongolia University of Science and Technology, No. 2010NC037*

PDF

668

摘要：

背景：ITK主要提供医学图像处理、分割与配准算法，但其缺少可视化的功能，缺乏灵活实用的用户界面，VTK提供了丰富的医学影像处理与分析工具，具有强大的图形处理和可视化功能。
目的：利用以前的确诊病例和医生的诊断经验以及患者的相关病史，对确诊的医学影像资源进行管理，归档，并检索，以减少人工干预，提高图像的查全率和查准率。
方法：以视觉感知机制为基础，在ITK平台上进行图像平滑去噪和分割的预处理过程，利用Tamura算法完成纹理特征提取，最后通过实验采集、计算数据，完成对比分析。
结果与结论：基于图像分割的Tamura纹理特征算法在基于图像纹理检索应用上便于相似性度量，进而可提高检索的准确率。

关键词: 图像分割, 基于内容的医学图像检索, 平滑去噪, Tamura, ITK

Abstract:

BACKGROUND: Insight segmentation and registration toolkit is the main provider of medical image processing, segmentation and registration algorithm, but, it lacks of visual function and flexible and practical user interface. Visualization toolkit provides a wealth of medical image processing and analysis tools, with the powerful graphics and visualization capabilities.
OBJECTIVE: To manage, archive and retrieve the diagnosed medical imaging resource by using the previously confirmed cases, diagnose experience and relevant medical history in order to reduce manual intervention and improve the image recall rate and precision rate.
METHODS: The images were preformed with smoothing, de-noising and segmentation pretreatment process on the insight segmentation and registration toolkit platform based on visual perception mechanism, and then texture feature was extracted by Tamura algorithm. Finally, the comparative analysis was completed through experiment data collection and calculation.
RESULTS AND CONCLUSION: The experiment has proved that the method based on the application of texture feature retrieval can facilitate the similarity measurement and improve the retrieval accuracy.
 

中图分类号: