background

Projects





Multimedia Retrieval & Data Visualization


Multimedia Retrieval includes Text, Audio, Image & Video Retrieval. Data Visualization includes the various visualization techniques: Curve plot, Stem plot, Scatter plot; 2D Compass plot, 2D plot, 3D Surface plot, 3D line plot; Data visualization using Principal Component Analysis (PCA) and Generative Topographic Mapping (GTM)    more

Face Recognition


We have developed advanced face recognition technologies, such as the Gabor-Fisher classifier (GFC), the dimensionality increasing framework (DIF), the Kernel Fisher Analysis (KFA). The DIF for example achieves the best performance on the government organized Face Recognition Grand Challenge (FRGC) competition.    more

Face Detection


A novel Bayesian discriminating features (BDF) method is proposed for multiple frontal face detection. The novelty of this method comes from the integration of the discriminating feature analysis of the input image, the statistical modeling of face and nonface classes, and the Bayes classifier for multiple frontal face detection.   more

Eye Detection


FLBP, LQP and FLQP are proposed for computing features. Three Clustering-based discriminant analysis (CDA) models are proposed to address the problem that the Fisher linear discriminant may not be able to extract adequate features for satisfactory performance, especially for two class problems, such as eye detection. Also, a novel efficient support vector machine is proposed to achieve the real-time speed.   more

Iris Recognition


A new iris recognition method based on a robust iris segmentation approach is presented. The robust iris segmentation approach applies power-law transformations for more accurate detection of the pupil region, which significantly reduces the candidate limbic boundary search space for increasing detection accuracy and efficiency.   more

Image Search


Many image search methods are proposed, such as Gabor-PHOG, Wigner-based LBP, hierarchical HITS model, locally linear KNN classifier. These methods are applied on action recognition, scene recognition and object recognition.   more

Innovative Color Models & Color Feature Extraction Methods


Numerous innovative color models and color feature extraction methods are developed for improving pattern recognition performance.    more

New Similarity Measures


Theoretical analysis of why the PCA based feature extraction methods favor the whitened cosine similarity measure, while the discriminant analysis based feature extraction methods care for the cosine similarity measure. New similarity measures that improve upon the traditional ones are proposed, such as the PRM Whitened Cosine (PWC) similarity measure, the Within-class Whitened Cosine (WWC) similarity measure, and the new similarity measure that integrate the angular measure and Lp norm.   more

Evolutionary Pursuit


The Evolutionary Pursuit (EP) method, which searches all the possible subspaces to derive the best features for both classification and generalization performance, performs better than other popular feature extraction methods, such as the PCA, the FLD, and the Independent Component Analysis (ICA) methods.    read the paper