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 ›› 2013, Vol. 39 ›› Issue (4): 226-229,233. doi: 10.3969/j.issn.1000-3428.2013.04.052

• 人工智能及识别技术 • 上一篇    下一篇

基于得分的近红外线与可见光图像融合算法

潘 磊1,尹义龙1,李徐周2   

  1. (1. 山东大学计算机科学与技术学院,济南 250101;2. 山东青年政治学院,济南 250103)
  • 收稿日期:2012-06-05 出版日期:2013-04-15 发布日期:2013-04-12
  • 作者简介:潘 磊(1985-),男,硕士,主研方向:人脸识别;尹义龙,教授;李徐周,讲师
  • 基金资助:
    山东省高等学校科技计划基金资助项目(J11LG28);高等学校博士学科点专项科研基金资助项目(20100131110021)

Near Infrared Ray and Visible Light Image Fusion Algorithm Based on Score

PAN Lei 1, YIN Yi-long 1, LI Xu-zho 2   

  1. (1. School of Computer Science and Technology, Shandong University, Jinan 250101, China; 2. Shandong Youth University of Political Science, Jinan 250103, China)
  • Received:2012-06-05 Online:2013-04-15 Published:2013-04-12

摘要: 近红外线图像与可见光图像相比,会出现模糊、轮廓不全等问题,对识别性能造成影响。为此,提出一种基于得分的近红外线与可见光图像融合算法,采用特征脸算法对2种模态的人脸样本进行训练学习,在获得2种图像各自的匹配得分后,进行基于得分的融合,以获得最终得分。实验结果表明,该算法的识别效果优于各单一条件下的识别性能,等错误率在CASIA HFB数据库上由4.65%降至3.80%,在HITSZ Lab1数据库上由0.55%降至0.38%。

关键词: 模式识别, 近红外线, 人脸识别, 可见光, 特征脸

Abstract: Compared with the Visible Light(VL) image, the Near Infrared Ray(NIR) image leads to fuzzy and insufficiency outline of face which is impact on the performance of face recognition. A new NIR and VL fusion algorithm based on score is proposed. Matching scores of NIR and VL face images are got by eigenface algorithm, and is to fuse the scores to get the final score. Experimental results show the effectiveness of the proposed method. The Equal Error Rate(EER) on CASIA HFB database is reduced from 4.65% to 3.80% and it is reduced from 0.55% to 0.38% on HITSZ Lab1 database.

Key words: mode recognition, Near Infrared Ray(NIR), face recognition, Visible Light(VL), eigenface

中图分类号: