Academic
Publications
Video-Based Face Recognition Using Adaptive Hidden Markov Models

Video-Based Face Recognition Using Adaptive Hidden Markov Models,10.1109/CVPR.2003.1211373,Xiaoming Liu,Tsuhan Cheng

Video-Based Face Recognition Using Adaptive Hidden Markov Models   (Citations: 101)
BibTex | RIS | RefWorks Download
While traditional face recognition is typically based on still images, face recognition from video sequences has become popular recently. In this paper, we propose to use adaptive Hidden Markov Models (HMM) to perform video- based face recognition. During the training process, the statistics of training video sequences of each subject, and the temporal dynamics, are learned by an HMM. During the recognition process, the temporal characteristics of the test video sequence are analyzed over time by the HMM corresponding to each subject. The likelihood scores provided by the HMMs are compared, and the highest score provides the identity of the test video sequence. Furthermore, with unsupervised learning, each HMM is adapted with the test video sequence, which results in better modeling over time. Based on extensive experiments with various databases, we show that the proposed algorithm provides better performance than using majority voting of image-based recognition results.
Conference: Computer Vision and Pattern Recognition - CVPR , vol. 1, pp. 340-345, 2003
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
    • ...Tracking can also be combined with either detection [1], [5], [44] or recognition [15], [28], [29], [36], [39], [47]...
    • ...In [29], the hidden Markov model is used for video-based face recognition...
    • ...Previous studies have shown theeffectivenessofsimultaneoustrackingandrecognitionon certain cases [15], [47], [36], [29], [39], [35]...

    Xue Meiet al. Robust Visual Tracking and Vehicle Classification via Sparse Represent...

    • ...2003) and hidden Markov model (HMM) (Liu and Chen 2003) are also used for video-based face recognition...

    M. Balasubramanianet al. Video-based person recognition using fovea intensity comparison code

    • ...Many theories can be adopted in emotion estimation such as Bayesian theory, Hidden Markov Model[4], Fuzzy logic theory etc...

    Wei Zhouet al. Towards 3D Communications: Real Time Emotion Driven 3D Virtual Facial ...

    • ...Recent work has shown success in recognizing faces by exploring temporal information [20]–[22]; however, the data modality is still in two dimensions...
    • ...Inspired by previous research on face recognition [20], [34], [35] using hidden Markov models (HMMs), we propose to use a 3-D surface primitive feature with HMM to investigate the 3-D spatiotemporal facial behavior in the 4-D domain...
    • ...We conducted experiments and compared the results with the following: 1) a 2-D-image-based (e.g., Gabor-waveletbased approach [36]); 2) static 3-D-model-based approaches using linear discriminant analysis (LDA) [37], [38] and principal component analysis (PCA); and 3) a video-based adaptive HMM (A-HMM) method [20] for face recognition...
    • ...Motivated by the recent success of using HMMs for 2-D-video-based face recognition [20], we propose an ST-HMM to analyze 3-D facial motions in order to learn individual facial characteristics...
    • ...C. Face Recognition: Comparison of 4-D and 2-D Videos [20]...
    • ...In order to further evaluate our 4-D databased face recognition approach, we compared our approach with the existing work reported by Liu and Chen [20], which applied a so-called A-HMM for 2-D-video-based face recognition...
    • ...COMPARISON WITH LIU AND CHEN’S METHOD [20] USING THE...
    • ...Following the algorithm described in [20], we cropped the facial regions from all the frames of 2-D video sequences...
    • ...According to Liu and Chen [20], weighting factors α and β are chosen to be 0.5 and 0.3, respectively...
    • ...The result shows that our 4-D databased approach outperforms the 2-D-video-based approach [20]...
    • ...However, the 2-D image patches used in [20] are sensitive to illumination, scale, and expression...
    • ...Our 3-D dynamic geometric-modelbased approach has shown a significant improvement over the 2-D-video-based approach [20] in the situation of expression-independent face recognition...
    • ...3) Our observation vectors are obtained by LDA transformation, while Liu and Chen [20] used the PCA dimensionality reduction method...

    Yi Sunet al. Tracking Vertex Flow and Model Adaptation for Three-Dimensional Spatio...

    • ...Recent work and efforts on human recognition have shown that the human behavior (e.g. facial dynamics [11]) and motion (e.g...

    Anastasios Drosouet al. Event-based unobtrusive authentication using multi-view image sequence...

Sort by: