计算机工程与应用 ›› 2014, Vol. 50 ›› Issue (7): 149-153.

• 图形图像处理 • 上一篇    下一篇

相关滤波器和2DPCA相融合的变光照人脸识别

吴曦德1,2,周庆标1   

  1. 1.浙江工业职业技术学院,浙江 绍兴 312000
    2.华中科技大学 计算机工程学院,武汉 430030
  • 出版日期:2014-04-01 发布日期:2014-04-25

Variable illumination face recognition based on correlation filter and 2DPCA

WU Xide1,2, ZHOU Qingbiao1   

  1. 1.Zhejiang Industry Polytechnic College, Shaoxing, Zhejiang 312000, China
    2.School of Computer Engineering, Huazhong University of Science & Technology, Wuhan 430030, China
  • Online:2014-04-01 Published:2014-04-25

摘要: 为解决变光照下人脸识别的识别率低问题,提出一种最佳相关滤波和2DPCA相融合的光照人脸识别方法。通过采用特定类2DPCA重构人脸图像,生成一对相关滤波器;测试人脸图像通过相关性滤器将投影到二维子空间中,并根据预先设定的峰旁瓣比阈值进行人脸识别;最后采用PIE和YaleB人脸库进行仿真实验。相比其他人脸识别方法,该方法获得了更高的人脸识别率,鲁棒性更强。

关键词: 人脸识别, 光照变化, 相关滤波器, 最佳投影图像

Abstract: In order to improve the recognition rate, a novel variable illumination face recognition method is proposed based on correlation filter and 2DPCA. Firstly, a pair of correlation filters is generated from input image projected by class specific 2DPCA, and then the test face images are projected onto a 2D subspace by correlation filter, and face image is recognized by threshold of PSR, finally, the simulation experiments are carried out to test performance of the proposed method on YaleB and PIE image database. The proposed method obtains higher recognition rate and more robust than the other image recognition methods.

Key words: face recognition, illumination, correlation filter, optimal projection