Y
Yap-Peng Tan
Researcher at Nanyang Technological University
Publications - 296
Citations - 9430
Yap-Peng Tan is an academic researcher from Nanyang Technological University. The author has contributed to research in topics: Facial recognition system & Feature extraction. The author has an hindex of 47, co-authored 290 publications receiving 8521 citations. Previous affiliations of Yap-Peng Tan include Fudan University & Intel.
Papers
More filters
Proceedings ArticleDOI
Frame-level quantization control for perceptual quality constrained H.264/AVC video coding
Zhenzhong Chen,Yap-Peng Tan +1 more
TL;DR: A frame-level quantization control approach for perceptual quality constrained video coding which aims to compress video at a certain perceptual quality level is presented and a general learning-based framework in which different video quality measures can be adopted is developed.
Proceedings ArticleDOI
View recognition of human gait sequences in videos
Jiwen Lu,Yap-Peng Tan +1 more
TL;DR: A new adaptive discriminant analysis (ADA) method is proposed by imposing large penalties on interclass samples with small differences and small penalties on those samples with large differences simultaneously, such that the discriminating power of the extracted features can be boosted for view recognition.
Proceedings ArticleDOI
Respiratory rate estimation via simultaneously tracking and segmentation
TL;DR: A novel strategy of simultaneously tracking and segmentation is proposed for human respiratory rate estimation from thermal infrared, which can be applicable to contact-free polygraphy, airport health screening and patient monitoring system and also utilized for any applications regarding the automatic detection and tracking of pixels experiencing periodic intensity variation.
Proceedings ArticleDOI
Efficient video motion estimation using dual-cross search algorithms
Xuan-Quang Banh,Yap-Peng Tan +1 more
TL;DR: Two new and efficient block-matching algorithms for fast motion estimation which is commonly used in motion-compensated video compression are presented and outperform conventional algorithms in terms of both peak signal-to-noise ratio and the number of search points evaluated.
Journal ArticleDOI
Generalized and Discriminative Collaborative Representation for Multiclass Classification
TL;DR: The proposed DRC method has the following three desirable properties: discriminability: DRC can leverage the label information of training data and is supervised in both representation and classification, thus improving the discriminative ability of the representation vector; efficiency: it has a closed-form solution and is efficient in computing the representation vectors and performing classification; and theory: it also has theoretical guarantees for classification.