Zhenjun Han

Visual object tracking via sample-based Adaptive Sparse Representation (AdaSR)

When appearance variation of object and its background, partial occlusion or deterioration in object images occurs, most existing visual tracking methods tend to fail in tracking the target. To address this problem, this paper proposes a new approach for visual object tracking based on Sample-Based Adaptive Sparse Representation (AdaSR), which ensures that the tracked object is adaptively and compactly expressed with predefined samples. First, the Sample-Based Sparse Representation, which selects a subset of samples as a basis for object representation by exploiting L1-norm minimization, improves the representation adaptation to partial occlusion for tracking. Second, to keep the temporal consistency and adaptation to appearance variation and deterioration in object images during the tracking process, the objects Sample-Based Sparse Representation is adaptively evaluated based on a Kalman filter, obtaining the AdaSR. Finally, the candidate holding the most similar Sample-Based Sparse Representation to the AdaSR of the tracked object will be regarded as the instantaneous tracking result. In addition, we can easily extend the AdaSR for multi-object tracking by integrating the sample set of each tracked object (named Common Sample-Based Adaptive Sparse Representation Analysis (AdaSRA)). AdaSRA fully analyses Adaptive Sparse Representation similarity for object classification. Our experiments on public datasets show state-of-the-art results, which are better than those of several representative tracking methods.

Visual abnormal behavior detection based on trajectory sparse reconstruction analysis

Abnormal behavior detection has been one of the most important research branches in intelligent video content analysis. In this paper, we propose a novel abnormal behavior detection approach by introducing trajectory sparse reconstruction analysis (SRA). Given a video scenario, we collect trajectories of normal behaviors and extract the control point features of cubic B-spline curves to construct a normal dictionary set, which is further divided into Route sets. On the dictionary set, sparse reconstruction coefficients and residuals of a test trajectory to the Route sets can be calculated with SRA. The minimal residual is used to classify the test behavior into a normal behavior or an abnormal one. SRA is solved by L1-norm minimization, leading to that a few of dictionary samples are used when reconstructing a behavior trajectory, which guarantees that the proposed approach is valid even when the dictionary set is very small. Experimental results with comparisons show that the proposed approach improves the state-of-the-art.

Online feature evaluation for object tracking using Kalman Filter

An online feature evaluation method for visual object tracking is put forward in this paper. Firstly, a combined feature set is built using color histogram (HC) bins and gradient orientation histogram (HOG) bins considering the color and contour representation of an object respectively. Then a novel method is proposed to evaluate the features¿ weights in a tracking process using Kalman Filter, which is used to comprise the inter-frame predication and single-frame measurement of features¿ discriminative power. In this way, we extend the traditional filter framework from modeling motion states to modeling feature evaluation. Experiments show this method can greatly improve the tracking stabilization when objects go across complex backgrounds.

Abnormal Behavior Detection via Sparse Reconstruction Analysis of Trajectory

This paper proposes a new method for abnormal behavior detection in surveillance videos via sparse reconstruction analysis. The motion trajectories of objects are firstly defined as fixed-length parametric vectors based on approximating cubic B-spline curves. Then the vectors are classified as behavior patterns and finally distinguished between normal and abnormal behaviors based on sparse reconstruction analysis, in which a classifier is constructed with sparse linear reconstruction coefficients by computing L1-norm minimization and sparse reconstruction residuals learning from labeled training samples. Experimental results on public dataset show the effectiveness of the proposed approach.

Human Detection in Images via Piecewise Linear Support Vector Machines

Human detection in images is challenged by the view and posture variation problem. In this paper, we propose a piecewise linear support vector machine (PL-SVM) method to tackle this problem. The motivation is to exploit the piecewise discriminative function to construct a nonlinear classification boundary that can discriminate multiview and multiposture human bodies from the backgrounds in a high-dimensional feature space. A PL-SVM training is designed as an iterative procedure of feature space division and linear SVM training, aiming at the margin maximization of local linear SVMs. Each piecewise SVM model is responsible for a subspace, corresponding to a human cluster of a special view or posture. In the PL-SVM, a cascaded detector is proposed with block orientation features and a histogram of oriented gradient features. Extensive experiments show that compared with several recent SVM methods, our method reaches the state of the art in both detection accuracy and computational efficiency, and it performs best when dealing with low-resolution human regions in clutter backgrounds.

Combined feature evaluation for adaptive visual object tracking

Existing visual tracking methods are challenged by object and background appearance variations, which often occur in a long duration tracking. In this paper, we propose a combined feature evaluation approach in filter frameworks for adaptive object tracking. First, a feature set is constructed by combining color histogram (HC) and gradient orientation histogram (HOG), which gives a representation of both color and contour. Then, to adapt to the appearance changes of the object and its background, these features are assigned with different confidences adaptively to make the features with higher discriminative ability play more important roles in the instantaneous tracking. To keep the temporal consistency, the feature confidences are evaluated based on Kalman and Particle filters. Experiments and comparisons demonstrate that object tracking with evaluated features have good performance even when objects go across complex backgrounds.

Corrosion behavior of Ti-6Al-4V alloy welded by scanning electron beam

The corrosion behavior of Ti-6Al-4V alloy welded by SEE (scanning electron beam) is described in this paper. Solidification macrostructure of weld was transformed from 650-mu m columnar crystals to 100-mu m equiaxed crystals by SEE. Anodic polarization behavior of the weld was different following the change in solidification macrostructure. Anodic polarization curve of 150-mu m equiaxed-grain weld coincided with that of base metal, which indicated a best state of equal corrosion tendency between the base and the weld metal. Alloy element homogeneity in the equiaxed solidification structures made the weld more corrosion resistant than the columnar solidification structures. Fine microstructure and a small quantity of dislocations have a good effect on uniform corrosion resistance of the weld, but coarse microstructure, a spinodal decomposition of the metastable beta-phase and a large quantity of dislocations were related to the inferior resistance either to uniform corrosion or to local corrosion

A new segment-based algorithm for stereo matching

A new segment-based algorithm for stereo matching is put forward in this paper. Firstly, the reference image is segmented by mean shift algorithm and an adaptive window matching method using histogram of orientation gradient is presented. Secondly, a set of reliable pixels is constructed consistent between the left-right initial disparity maps, which can lead to an actual disparity plane. Thirdly, the improved hierarchical clustering algorithm is applied to merge the neighboring segment regions. Angle and distance are used together to measure whether the two planes are the same or not, which will reduce the calculation complexity. Finally, we derive the plane set in each region and the final disparity map is produced. Experiments verify the advantages.

Dealloying characterizations of Cu-Al alloy in marine environment

chinese acad sci, inst corros & protect met, state key lab corros & protect, shenyang 110015, peoples r china. shenyang polytech coll, shenyang, peoples r china.;han, z (reprint author), chinese acad sci, inst corros & protect met, state key lab corros & protect, shenyang 110015, peoples r china

Robust weld line detection with cross structured light and Hidden Markov Model

In this paper, we propose a vision-based weld line detection system with cross structured light (CSL) and Hidden Markov Model (HMM). Our consideration is that the laser stripes projected by CSL can reflect the convex shapes of weld lines, capture horizontal and vertical weld lines simultaneously and are insensitive to illumination changes. The Hidden Markov Model filters out various noises on the weldment surface and then detects laser stripes precisely, which guarantees the accuracy of weld line location. Experiments on practical and simulated datasets show that the results of our approach have great improvement to classical Hough transform based method.