An Approach to Improving Single Sample Face Recognition Using High Confident Tracking Trajectories
In this paper, single sample face recognition (SSFR) problem is addressed by introducing an adaptive biometric system within a modular architecture where one detector per target individual is proposed. For each detector, a face model is generated with the gallery face image and updated overtime. Sequential Karhunen-Loeve technique is applied to update the face model using representative face captures which are selected from the operational data by using reliable tracking trajectories. This process helps to induce intra-class variation of face appearance and improve representativeness of the face models. The effectiveness of the proposed method is detailed in security surveillance and user authentication using Chokepoint and FIA datasets in SSFR setting.
This research is supported by Academic Research Fund and Research Incentive Grant, Athabasca University, and NSERC, Canada.
- 8.Bashbagi, S., Granger, E., Sabourin, R., Bilodeau, G.-A.: Watch-list screening using ensembles based on multiple face representations. In: ICPR, Stockholm, Sweden (2014)Google Scholar
- 9.Wong, Y., Chen, S., Mau, S., Sanderson, C., Lovell, B.C.: Patch-based probabilistic image quality assessment for face selection and improved video-based face recognition. In: CVPRW, Colorado, USA (2011)Google Scholar
- 10.Goh, R., Liu, L., Liu, X., Chen, T.: The CMU face in action database. In: AMFG, Beijing, China (2005)Google Scholar