Hanqing Lu

Face detection using improved LBP under Bayesian framework

In this paper, we present a novel face detection approach using improved local binary patterns (ILBP) as facial representation. ILBP feature is an improvement of LBP feature that considers both local shape and texture information instead of raw grayscale information and it is robust to illumination variation. We model the face and non-face class using multivariable Gaussian model and classify them under Bayesian framework. Extensive experiments show that the proposed method has an encouraging performance.

Solving the small sample size problem of LDA

The small sample size problem is often encountered in pattern recognition. It results in the singularity of the within-class scattering matrix S/sub w/ in linear discriminant analysis (LDA). Different methods have been proposed to solve this problem in face recognition literature. Some methods reduce the dimension of the original sample space and hence unavoidably remove the null space of S/sub w/, which has been demonstrated to contain considerable discriminative information; whereas other methods suffer from the computational problem. In this paper, we propose a new method making use of the null space of S/sub w/ effectively and solve the small sample size problem of LDA. We compare our method with several well-known methods, and demonstrate the efficiency of our method.

Robust Structured Subspace Learning for Data Representation

To uncover an appropriate latent subspace for data representation, in this paper we propose a novel Robust Structured Subspace Learning (RSSL) algorithm by integrating image understanding and feature learning into a joint learning framework. The learned subspace is adopted as an intermediate space to reduce the semantic gap between the low-level visual features and the high-level semantics. To guarantee the subspace to be compact and discriminative, the intrinsic geometric structure of data, and the local and global structural consistencies over labels are exploited simultaneously in the proposed algorithm. Besides, we adopt the `2;1-norm for the formulations of loss function and regularization respectively to make our algorithm robust to the outliers and noise. An efficient algorithm is designed to solve the proposed optimization problem. It is noted that the proposed framework is a general one which can leverage several well-known algorithms as special cases and elucidate their intrinsic relationships. To validate the effectiveness of the proposed method, extensive experiments are conducted on diversity datasets for different image understanding tasks, i.e., image tagging, clustering, and classification, and the more encouraging results are achieved compared with some state-of-the-art approaches.

Clustering-Guided Sparse Structural Learning for Unsupervised Feature Selection

Many pattern analysis and data mining problems have witnessed high-dimensional data represented by a large number of features, which are often redundant and noisy. Feature selection is one main technique for dimensionality reduction that involves identifying a subset of the most useful features. In this paper, a novel unsupervised feature selection algorithm, named clustering-guided sparse structural learning (CGSSL), is proposed by integrating cluster analysis and sparse structural analysis into a joint framework and experimentally evaluated. Nonnegative spectral clustering is developed to learn more accurate cluster labels of the input samples, which guide feature selection simultaneously. Meanwhile, the cluster labels are also predicted by exploiting the hidden structure shared by different features, which can uncover feature correlations to make the results more reliable. Row-wise sparse models are leveraged to make the proposed model suitable for feature selection. To optimize the proposed formulation, we propose an efficient iterative algorithm. Finally, extensive experiments are conducted on 12 diverse benchmarks, including face data, handwritten digit data, document data, and biomedical data. The encouraging experimental results in comparison with several representative algorithms and the theoretical analysis demonstrate the efficiency and effectiveness of the proposed algorithm for feature selection.

Using Webcast Text for Semantic Event Detection in Broadcast Sports Video

Sports video semantic event detection is essential for sports video summarization and retrieval. Extensive research efforts have been devoted to this area in recent years. However, the existing sports video event detection approaches heavily rely on either video content itself, which face the difficulty of high-level semantic information extraction from video content using computer vision and image processing techniques, or manually generated video ontology, which is domain specific and difficult to be automatically aligned with the video content. In this paper, we present a novel approach for sports video semantic event detection based on analysis and alignment of Webcast text and broadcast video. Webcast text is a text broadcast channel for sports game which is co-produced with the broadcast video and is easily obtained from the Web. We first analyze Webcast text to cluster and detect text events in an unsupervised way using probabilistic latent semantic analysis (pLSA). Based on the detected text event and video structure analysis, we employ a conditional random field model (CRFM) to align text event and video event by detecting event moment and event boundary in the video. Incorporation of Webcast text into sports video analysis significantly facilitates sports video semantic event detection. We conducted experiments on 33 hours of soccer and basketball games for Webcast analysis, broadcast video analysis and text/video semantic alignment. The results are encouraging and compared with the manually labeled ground truth.

A Real-Time Hand Gesture Recognition Method

This paper presents a real-time hand gesture recognition method. Hand regions are located by combining adaptive skin color detection and motion detection. Histograms of oriented gradients (HOG) are used to describe hand image. Extracted HOG features are projected into low-dimensional subspace using PCA-LDA (Principle Component Analysis and Linear Discriminant Analysis). And then distances between the projected features and each gesture class center are calculated. Finally the recognition results are obtained by the nearest neighbor method. Experimental results showed that the proposed method gained detection rate up to 91% with real-time performance.

Letter: Supervised kernel locality preserving projections for face recognition

Subspace analysis is an effective approach for face recognition. Finding a suitable low-dimensional subspace is a key step of subspace analysis, for it has a direct effect on recognition performance. In this paper, a novel subspace method, named supervised kernel locality preserving projections (SKLPP), is proposed for face recognition, in which geometric relations are preserved according to prior class-label information and complex nonlinear variations of real face images are represented by nonlinear kernel mapping. SKLPP cannot only gain a perfect approximation of face manifold, but also enhance local within-class relations. Experimental results show that the proposed method can improve face recognition performance.

A robust boosting tracker with minimum error bound in a co-training framework

The varying object appearance and unlabeled data from new frames are always the challenging problem in object tracking. Recently machine learning methods are widely applied to tracking, and some online and semi-supervised algorithms are developed to handle these difficulties. In this paper, we consider tracking as a classification problem and present a novel tracking method based on boosting in a co-training framework. The proposed tracker can be online updated and boosted with multi-view weak hypothesis. The most important contribution of this paper is that we find a boosting error upper bound in a co-training framework to guide the novel tracker construction. In theory, the proposed tracking method is proved to minimize this error bound. In experiments, the accuracy rate of foreground/ background classification and the tracking results are both served as evaluation metrics. Experimental results show good performance of proposed novel tracker on challenging sequences.

Saliency Cuts: An automatic approach to object segmentation

Interactive graph cuts are widely used in object segmentation but with some disadvantages: 1) Manual interactions may cause inaccurate or even incorrect segmentation results and involve more interactions especially for novices. 2) In some situations, the manual interactions are infeasible. To overcome these disadvantages, we propose a novel approach, namely Saliency cuts, to segment object from background automatically. By exploring the effects of labels to graph cuts, the so called ldquoprofessional labelsrdquo is introduced to evaluate labels. With the aid of saliency detection, a multiresolution framework is designed to provide ldquoprofessional labelsrdquo automatically and implement object segmentation using graph cuts. The experiments demonstrate the promising performance of Saliency cuts.

Adaptive rate control for H.264

This paper presents a rate control scheme for H.264 by introducing the concept of basic unit and a linear prediction model. The basic unit can be a macroblock (MB), a slice, or a frame. It can be used to obtain a trade-off between the overall coding efficiency and the bits fluctuation. The linear model is used to solve the chicken and egg dilemma existing in the rate control of H.264. Both constant bit rate (CBR) and variable bit rate (VBR) cases are studied. Our scheme has been adopted by H.264.