吉林大学学报(信息科学版)

• 论文 • 上一篇    下一篇

基于引导滤波的指节纹图像增强算法研究

李温温1,2,刘富1,姜守坤1   

  1. 1. 吉林大学通信工程学院,长春130022; 2. 白城师范学院机械工程学院,吉林白城137000
  • 收稿日期:2017-12-18 出版日期:2018-07-24
  • 作者简介:李温温( 1984— ) ,女,山西长治人,白城师范学院副教授,吉林大学博士研究生,主要从事模式识别与智能控制研究,( Tel) 86-13634360943( E-mail) liufu@ jlu. edu. cn; 刘富( 1968— ) ,男,长春人,吉林大学教授,博士生导师,主要从事计算机视觉与模式识别研究,( Tel) 86-13610708679( E-mail) liufu@ jlu. edu. cn。
  • 基金资助:
    吉林省教育厅“十三五”科学技术基金资助项目( JJKH20170003KJ)

Research on Image Enhancement Algorithm of Finger Knuckle Print Based on Guided Image Filtering

LI Wenwen1,2,LIU Fu1,JIANG Shoukun1   

  1. 1. College of Communications Engineering,Jilin University,Changchun 130022,China;2. College of Mechanical Engineering,Baicheng Normal University,Baicheng 137000,China
  • Received:2017-12-18 Online:2018-07-24

摘要: 在利用指节纹进行身份识别时,为提高图像的质量,避免光照不均匀和环境噪声对特征提取带来的影响,需要在图像预处理阶段进行增强处理。针对指节纹图像的特点,设计了基于梯度图像引导滤波的图像增强算法,并在指背关节纹数据库中进行了验证。结果表明,该图像增强算法可避免光照不均带来的影响,同时突出细节特征,适用于指节纹图像的预处理。最后进行的定量分析证明,经过本算法的增强处理后,图像将保留更多的细节信息特征,有利于后续特征提取和提高图像识别的准确性。

关键词: 图像增强, 信息熵, 指节纹, 引导滤波

Abstract: In order to improve the image quality of images and degrade the effect of uneven illumination ambient noise in the identity recognition based on finger knuckle print,it is necessary to enhance the image in the process of image preprocessing. A new method of finger knuckle print image enhancement is proposed based on guided image filtering,due to abundant texture features of finger knuckle print image. And effectiveness is validated through the experiment in the database of finger knuckle print. The results show that the effect of uneven illumination can be avoided and details of texture can be highlighted by using the method of image enhancement.Conclusions were given by quantitative study,and more detail information are preserved after the imageenhancement,which contribute to the accuracy of subsequent feature extraction and identification.

Key words: finger knuckle print, information entropy, image enhancement, guided image filtering

中图分类号: 

  • TP391