DOI QR코드

DOI QR Code

An Improved Joint Bayesian Method using Mirror Image's Features

미러영상 특징을 이용한 Joint Bayesian 개선 방법론

  • Han, Sunghyu (School of Liberal Arts, Korea University of Technology and Education) ;
  • Ahn, Jung-Ho (Division of Computer Media Information Engineering, Kangnam University)
  • Received : 2015.06.22
  • Accepted : 2015.10.10
  • Published : 2015.10.31

Abstract

The Joint Bayesian[1] method was published in 2012. Since then, it has been used for binary classification in almost all state-of-the-art face recognition methods. However, no improved methods have been published so far except 2D-JB[2]. In this paper we propose an improved version of the JB method that considers the features of both the given face image and its mirror image. In pattern classification, it is very likely to make a mistake when the value of the decision function is close to the decision boundary or the threshold. By making the value of the decision function far from the decision boundary, the proposed method reduces the errors. The experimental results show that the proposed method outperforms the JB and 2D-JB methods by more than 1% in the challenging LFW DB. Many state-of-the-art methods required tons of training data to improve 1% in the LFW DB, but the proposed method can make it in an easy way.

Joint Bayesian 방법론[1]은 2012년 발표된 이후 최근까지 최고 성능을 보이는 거의 모든 얼굴인식 알고리즘에서 이진 분류를 위해 사용되고 있지만, 지금까지 이를 개선한 알고리즘은 2D-JB[2] 외에 거의 발표되지 않았다. 우리는 본 논문에서 주어진 얼굴 영상과 이를 좌우 반전시킨 미러 영상을 함께 고려함으로써 Joint Bayesian 방법론의 성능을 향상시킬 수 있는 방법론을 제안한다. 일반적인 패턴인식에서 결정함수 값이 결정경계 또는 임계치에 가까운 경우 오류가 발생할 확률이 높다. 제안한 방법론은 미러 영상의 특징을 이용하여 결정함수 값을 결정경계로부터 멀어지게 함으로써 오류를 줄이는 방법이다. 우리는 LFW DB를 이용한 실험을 통해 제안한 JB 개선 방법론이 기존 JB 방법론보다 1%이상 높은 인식률을 보임을 입증하였다. LFW DB를 이용한 기존 연구들에서 성능을 1% 높이기 위해 많은 학습데이터가 필요했음을 감안할 때, 제안한 방법론은 큰 의미가 있다고 볼 수 있다.

Keywords

References

  1. D. Chen, X. Cao, L. Wang, F. Wen, and J. Sun, "Bayesian Face Revisited: A Joint Formulation," Proc. ECCV, pp.566-579, 2012.
  2. S. Han, I.-Y. Lee, J.-H. Ahn, "Two-dimensional Joint Bayesian method for face verification", Journal of Information Processing Systems, in press, 2015.
  3. G. Huang, M. Ramesh, T. Berg, and E. Learned-Miller, "Labeled Faces in the Wild: A Database for Studying Face Recognition in Unconstrained Environments", University of Massachusetts, Amherst, Technical Report 07-49, Oct. 2007.
  4. X. Cao, D. Wipf, F. Wen, and G. Duan, "A practical Tranfer Learning Algorithm for Face Verification", Proc. ICCV, pp.3208-3215, Dec. 2013.
  5. D. Chen, X. Cao, F. Wen, and J. Sun, "Blessing of Dimensionality: High-dimensional Feature and Its Efficient Compression for Face Verification", Proc. CVPR, pp.3025-3032, June. 2013.
  6. Y. Sun, Y. Chen, X. Wang and X. Tang, "Deep Learning Face Representation by Joint Identity-Verification", Proc. NIPS, Dec. 2014.
  7. Y. Sun, X. Wang, and X. Tang, "Deep Learning Face Representation from Predicting 10,000 Classes", Proc. CVPR, pp.1891-1898, June, 2014.
  8. Y. Sun, X. Wang, and X. Tang, "Deeply learned face representations are sparse, selective, and robust", ArXiv:1412.1265, Dec. 2014.
  9. L. Wolf, T. Hassner, and Y. Taigman, "Effective Face Recognition by Combining Multiple Descriptors and Learned Background Statistics", IEEE TPAMI, Vol.33, No.10, 2011.
  10. C.M. Bishop, Pattern Recognition and Machine Learning, 1st ed., Springer, 2006.
  11. C. Cortes and V. Vapnik, "Support-Vector Networks", Machine Learning, Vol.20, No.3, pp.273-297, 1995 https://doi.org/10.1007/BF00994018
  12. T. Ahonen, A. Hadid, and M. Pietikainen, "Face description with local binary patterns: Application to face recognition", IEEE TPAMI, Vol.28, No.12, pp.2037-2041, 2006. https://doi.org/10.1109/TPAMI.2006.244
  13. X. Xiong and F.D. Torre, "Supervised Descent Method and its Application to Face Alignment", Proc. CVPR, pp.532-539, June, 2013.
  14. J.-H. Ahn, "An Improved RSR Method to Obtain the Sparse Projection Matrix", Journal of Digital Contents Society, Vol.16, No.4, pp.605-613, 2015. https://doi.org/10.9728/dcs.2015.16.4.605