TY - GEN
T1 - An improved eLBPH method for facial identity recognition
T2 - IEEE International Conference on Robotics and Biomimetics, IEEE-ROBIO 2015
AU - Xi, Xuanyang
AU - Qin, Zhengke
AU - Ding, Shuguang
AU - Qiao, Hong
N1 - Publisher Copyright:
© 2015 IEEE.
PY - 2015
Y1 - 2015
N2 - Face perception is one of the most important tasks in robot vision especially for service robots. The spatially enhanced local binary pattern histogram (eLBPH) method has been proved to be effective for facial image representation and analysis, but the expression factor isn't considered and the region-dividing method is rough. In this paper, inspired by the biological mechanism of human memory and face perception, we improve the eLBPH and propose a new method, expression-specific weighted local binary pattern histogram (EWLBPH). Accordingly, the new method introduces a semantic division process and an extended modulation process into the classical eLBPH. What's more, for the facial expression recognition, we propose a novel method which utilizes the convolutional deep belief network (CDBN) to extract discriminative information and represent them effectively. Finally, through experiments we verify the rationality and effectiveness of the improvement and two psychophysical findings.
AB - Face perception is one of the most important tasks in robot vision especially for service robots. The spatially enhanced local binary pattern histogram (eLBPH) method has been proved to be effective for facial image representation and analysis, but the expression factor isn't considered and the region-dividing method is rough. In this paper, inspired by the biological mechanism of human memory and face perception, we improve the eLBPH and propose a new method, expression-specific weighted local binary pattern histogram (EWLBPH). Accordingly, the new method introduces a semantic division process and an extended modulation process into the classical eLBPH. What's more, for the facial expression recognition, we propose a novel method which utilizes the convolutional deep belief network (CDBN) to extract discriminative information and represent them effectively. Finally, through experiments we verify the rationality and effectiveness of the improvement and two psychophysical findings.
UR - https://www.scopus.com/pages/publications/84964558139
U2 - 10.1109/ROBIO.2015.7418917
DO - 10.1109/ROBIO.2015.7418917
M3 - 会议稿件
AN - SCOPUS:84964558139
T3 - 2015 IEEE International Conference on Robotics and Biomimetics, IEEE-ROBIO 2015
SP - 1090
EP - 1095
BT - 2015 IEEE International Conference on Robotics and Biomimetics, IEEE-ROBIO 2015
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 6 December 2015 through 9 December 2015
ER -