Yanyun Qu

Discriminative Object Tracking via Sparse Representation and Online Dictionary Learning

We propose a robust tracking algorithm based on local sparse coding with discriminative dictionary learning and new keypoint matching schema. This algorithm consists of two parts: the local sparse coding with online updated discriminative dictionary for tracking (SOD part), and the keypoint matching refinement for enhancing the tracking performance (KP part). In the SOD part, the local image patches of the target object and background are represented by their sparse codes using an over-complete discriminative dictionary. Such discriminative dictionary, which encodes the information of both the foreground and the background, may provide more discriminative power. Furthermore, in order to adapt the dictionary to the variation of the foreground and background during the tracking, an online learning method is employed to update the dictionary. The KP part utilizes refined keypoint matching schema to improve the performance of the SOD. With the help of sparse representation and online updated discriminative dictionary, the KP part are more robust than the traditional method to reject the incorrect matches and eliminate the outliers. The proposed method is embedded into a Bayesian inference framework for visual tracking. Experimental results on several challenging video sequences demonstrate the effectiveness and robustness of our approach.

Unifying visual saliency with HOG feature learning for traffic sign detection

Traffic sign detection is important to a robotic vehicle that automatically drives on roads. In this paper, an efficient novel approach which is enlighten by the process of the human vision is proposed to achieve automatic traffic sign detection. The detection method combines bottom-up traffic sign saliency region with learning based top-down features of traffic sign guided search. The bottom-up stage could obtain saliency region of traffic sign and achieve computational parsimony using improved Model of Saliency-Based Visual Attention. The top-down stage searches traffic sign in these traffic sign saliency regions based on the feature Histogram of Oriented Gradient (HOG) and the classifier Support Vector Mechine (SVM). Experimental results show that, the proposed approach can achieve robustness to illumination, scale, pose, viewpoint change and even partial occlusion. The samllest detection size of traffic sign is 14×14, the average detection rate is 98.3% and the false positive rate is 5.09% in test image data set.

Discriminative subspace learning with sparse representation view-based model for robust visual tracking

In this paper, we propose a robust tracking algorithm to handle drifting problem. This algorithm consists of two parts: the first part is the G&D part that combines Generative model and Discriminative model for tracking, and the second part is the View-Based model for target appearance that corrects the result of the G&D part if necessary. In G&D part, we use the Maximum Margin Projection (MMP) to construct a graph model to preserve both local geometrical and discriminant structures of the data manifold in low dimensions. Therefore, such discriminative subspace combined with traditional generative subspace can benefit from both models. In addition, we address the problem of learning maximum margin projection under the Spectral Regression (SR) which results in significant savings in computational time. To further solve the drift, an online learned sparsely represented view-based model of the target is complementary to the G&D part. When the result of G&D part is unreliable, the view-based model can rectify the result in order to avoid drifting. Experimental results on several challenging video sequences demonstrate the effectiveness and robustness of our approach. HighlightsA tracker combines the generative and the discriminative model.Introduce maximum margin projection into object tracking.Transform MMP function to be solved under the Spectral Regression framework.Learning an incremental view-based model of the target with sparse representation.Experiments show our tracker to be more robust and stable than state-of-the-art methods.

Hyperspectral Image Restoration via Iteratively Regularized Weighted Schatten

Hyperspectral images (HSIs) are inevitably corrupted by mixture noise during their acquisition process, in which various kinds of noise, e.g., Gaussian noise, impulse noise, dead lines, and stripes, may exist concurrently. In this paper, mixture noise removal is well illustrated by the task of recovering the low-rank and sparse components of a given matrix, which is constructed by stacking vectorized HSI patches from all the bands at the same position. Instead of applying a traditional nuclear norm, a nonconvex low-rank regularizer, i.e., weighted Schatten p-norm (WSN), is introduced to not only give better approximation to the original low-rank assumption but also to consider the importance of different rank components. The resulted nonconvex low-rank matrix approximation (LRMA) model falls into the applicable scope of an augmented Lagrangian method, and its WSN minimization subproblem can be efficiently solved by generalized iterated shrinkage algorithm. Moreover, the proposed model is integrated into an iterative regularization schema to produce final results, leading to a completed HSI restoration framework. Extensive experimental testing on simulated and real data shows, both qualitatively and quantitatively, that the proposed method has achieved highly competent objective performance compared with several state-of-the-art HSI restoration methods.

p

In this paper, we focus on learning an adaptive appearance model robustly and effectively for object tracking. There are two important factors to affect object tracking, the one is how to represent the object using a discriminative appearance model, the other is how to update appearance model in an appropriate manner. In this paper, following the state-of-the-art tracking techniques which treat object tracking as a binary classification problem, we firstly employ a new gradient-based Histogram of Oriented Gradient (HOG) feature selection mechanism under Multiple Instance Learning (MIL) framework for constructing target appearance model, and then propose a novel optimization scheme to update such appearance model robustly. This is an unified framework that not only provides an efficient way of selecting the discriminative feature set which forms a powerful appearance model, but also updates appearance model in online MIL Boost manner which could achieve robust tracking overcoming the drifting problem. Experiments on several challenging video sequences demonstrate the effectiveness and robustness of our proposal.

-Norm Minimization

Bag-of-word (BOW) is used in many state-of-the-art methods of image classification, and it is especially suitable for multi-class classification. Many kinds of local features and classifiers are applicable for the BOW model. However, it is unclear which kind of local feature is the most distinctive and meanwhile robust, and which classifier can optimize classification performance. In this paper, we discuss the implementation choices in the BOW model. Further, we evaluate the influences of local features and classifiers on object and texture recognition methods in the framework of the BOW model. To evaluate the implementation choices, we use two popular datasets: the Xerox7 dataset and the UIUCTex dataset. Extensive experiments are carried out to compare the performance of different detectors, descriptors and classifiers in term of classification accuracy on the object category dataset and the texture dataset. We find that the combinational detector which combines the MSER detector with the Hessian-Laplacian detector is efficient to find discriminative regions. We also find that the SIFT descriptor performs better than the other descriptors for image classification, and that the SVM classifier with the EMD kernel is superior to other classifiers. More than that, we propose an EMD spatial kernel to encode the spatial information of local features. The EMD spatial kernel is implemented on the Xerox7 dataset, the 4-class VOC2006 dataset and the 4-class Caltech101 dataset. The experimental results show that the proposed kernel outperforms the EMD kernel which does not consider the spatial information in image classification.

Online multiple instance gradient feature selection for robust visual tracking

We investigate the scalable image classification problem with a large number of categories. Hierarchical visual data structures are helpful for improving the efficiency and performance of large-scale multi-class classification. We propose a novel image classification method based on learning hierarchical inter-class structures. Specifically, we first design a fast algorithm to compute the similarity metric between categories, based on which a visual tree is constructed by hierarchical spectral clustering. Using the learned visual tree, a test sample label is efficiently predicted by searching for the best path over the entire tree. The proposed method is extensively evaluated on the ILSVRC2010 and Caltech 256 benchmark datasets. The experimental results show that our method obtains significantly better category hierarchies than other state-of-the-art visual tree-based methods and, therefore, much more accurate classification.We investigate the scalable image classification problem with a large number of categories. Hierarchical visual data structures are helpful for improving the efficiency and performance of large-scale multi-class classification. We propose a novel image classification method based on learning hierarchical inter-class structures. Specifically, we first design a fast algorithm to compute the similarity metric between categories, based on which a visual tree is constructed by hierarchical spectral clustering. Using the learned visual tree, a test sample label is efficiently predicted by searching for the best path over the entire tree. The proposed method is extensively evaluated on the ILSVRC2010 and Caltech 256 benchmark datasets. The experimental results show that our method obtains significantly better category hierarchies than other state-of-the-art visual tree-based methods and, therefore, much more accurate classification.

Evaluation of local features and classifiers in BOW model for image classification

It remains a challenge to simultaneously remove geometric distortion and space-time-varying blur in frames captured through a turbulent atmospheric medium. To solve, or at least reduce these effects, we propose a new scheme to recover a latent image from observed frames by integrating a new hybrid total variation model and deformation-guided spatial-temporal kernel regression. The proposed scheme first constructs a high-quality reference image from the observed frames using low-rank decomposition. Then, to generate an improved registered sequence, the reference image is iteratively optimized using a variational model containing the combined regularization of local and non-local total variations. The proposed optimization algorithm efficiently solves this model with convergence guarantee. Next, to reduce blur variation, deformation-guided spatial-temporal kernel regression is carried out to fuse the registered sequence into one image by introducing the concept of the near-stationary patch. Applying a blind deconvolution algorithm to the fused image produces the final output. Extensive experimental testing shows, both qualitatively and quantitatively, that the proposed method can effectively alleviate distortion, and blur and recover details of the original scene compared to the state-of-the-art methods.

Joint Hierarchical Category Structure Learning and Large-Scale Image Classification

In this paper, we address the multi-view subspace clustering problem. Our method utilizes the circulant algebra for tensor, which is constructed by stacking the subspace representation matrices of different views and then rotating, to capture the low rank tensor subspace so that the refinement of the view-specific subspaces can be achieved, as well as the high order correlations underlying multi-view data can be explored. By introducing a recently proposed tensor factorization, namely tensor-Singular Value Decomposition (t-SVD) (Kilmer et al. in SIAM J Matrix Anal Appl 34(1):148–172, 2013), we can impose a new type of low-rank tensor constraint on the rotated tensor to ensure the consensus among multiple views. Different from traditional unfolding based tensor norm, this low-rank tensor constraint has optimality properties similar to that of matrix rank derived from SVD, so the complementary information can be explored and propagated among all the views more thoroughly and effectively. The established model, called t-SVD based Multi-view Subspace Clustering (t-SVD-MSC), falls into the applicable scope of augmented Lagrangian method, and its minimization problem can be efficiently solved with theoretical convergence guarantee and relatively low computational complexity. Extensive experimental testing on eight challenging image datasets shows that the proposed method has achieved highly competent objective performance compared to several state-of-the-art multi-view clustering methods.

Removing Turbulence Effect via Hybrid Total Variation and Deformation-Guided Kernel Regression

In this paper, a novel approach to single image super-resolution based on the multi-kernel regression is presented. This approach focuses on learning the map between the space of high-resolution image patches and the space of blurred high-resolution image patches, which are the interpolation results generated from the corresponding low-resolution images. Kernel regression based super-resolution approaches are promising, but kernel selection is a critical problem. In order to avoid demanding and time-consuming cross validation for kernel selection, we propose multi-kernel regression (MKR) model for image Super-Resolution (SR). Considering the multi-kernel regression model is prohibited when the training data is large-scale, we further propose a prototype MKR algorithm which can reduce the computational complexity. Extensive experimental results demonstrate that our approach is effective and achieves a high quality performance in comparison with other super-resolution methods.