Zhan Song

Recent advances and trends in visual tracking: A review

The goal of this paper is to review the state-of-the-art progress on visual tracking methods, classify them into different categories, as well as identify future trends. Visual tracking is a fundamental task in many computer vision applications and has been well studied in the last decades. Although numerous approaches have been proposed, robust visual tracking remains a huge challenge. Difficulties in visual tracking can arise due to abrupt object motion, appearance pattern change, non-rigid object structures, occlusion and camera motion. In this paper, we first analyze the state-of-the-art feature descriptors which are used to represent the appearance of tracked objects. Then, we categorize the tracking progresses into three groups, provide detailed descriptions of representative methods in each group, and examine their positive and negative aspects. At last, we outline the future trends for visual tracking research.

Use of LCD Panel for Calibrating Structured-Light-Based Range Sensing System

Calibration is a crucial step in structured-light-based range sensing device. The step involves the determination of the intrinsic parameters of both the camera and the projector that constitute the device and the extrinsic parameters between the two instruments. The traditional solution requires the use of an external calibration object with an accurately measured pattern printed on it. This paper presents a calibration design that makes use of a liquid-crystal display (LCD) panel as the calibration object. The LCD panels planarity is of industrial grade and is thus dependable. The pattern shown on the LCD panel is programmable and is thus convenient to produce in high precision. We show that, with the design, the projector-and-camera system parameters can be calibrated with far fewer images with much higher accuracy. Extensive experiments are shown to illustrate the dramatic improvement in performance.

A Novel Recursive Bayesian Learning-Based Method for the Efficient and Accurate Segmentation of Video With Dynamic Background

Segmentation of video with dynamic background is an important research topic in image analysis and computer vision domains. In this paper, we present a novel recursive Bayesian learning-based method for the efficient and accurate segmentation of video with dynamic background. In the algorithm, each frame pixel is represented as the layered normal distributions which correspond to different background contents in the scene. The layers are associated with a confident term and only the layers satisfy the given confidence which will be updated via the recursive Bayesian estimation. This makes learning of background motion trajectories more accurate and efficient. To improve the segmentation quality, the coarse foreground is obtained via simple background subtraction first. Then, a local texture correlation operator is introduced to fill the vacancies and remove the fractional false foreground regions. Extensive experiments on a variety of public video datasets and comparisons with some classical and recent algorithms are used to demonstrate its improvements in both segmentation accuracy and efficiency.

Determining Both Surface Position and Orientation in Structured-Light-Based Sensing

Position and orientation profiles are two principal descriptions of shape in space. We describe how a structured light system, coupled with the illumination of a pseudorandom pattern and a suitable choice of feature points, can allow not only the position but also the orientation of individual surface elements to be determined independently. Unlike traditional designs which use the centroids of the illuminated pattern elements as the feature points, the proposed design uses the grid points between the pattern elements instead. The grid points have the essences that their positions in the image data are inert to the effect of perspective distortion, their individual extractions are not directly dependent on one another, and the grid points possess strong symmetry that can be exploited for their precise localization in the image data. Most importantly, the grid lines of the illuminated pattern that form the grid points can aid in determining surface normals. In this paper, we describe how each of the grid points can be labeled with a unique color code, what symmetry they possess and how the symmetry can be exploited for their precise localization at subpixel accuracy in the image data, and how 3D orientation in addition to 3D position can be determined at each of them. Both the position and orientation profiles can be determined with only a single pattern illumination and a single image capture.

An Accurate and Robust Strip-Edge-Based Structured Light Means for Shiny Surface Micromeasurement in 3-D

Three-dimensional measurement of shiny or reflective surface is a challenging issue for optical-based instrumentations. In this paper, we present a novel structured light approach for direct measurement of shiny target so as to skip the coating preprocedure. In comparison with traditional image-intensity-based structured light coding strategies like sinusoidal and line patterns, strip edges not raw image intensities are encoded in the illuminated patterns. With strip edges generally better preserved than individual image intensity in the image data in the presence of surface reflections, such a coding strategy is more robust. To remove the periodic ambiguity within strip patterns, traditional Gray code patterns are adopted. To localize the strip edges more precisely, both positive and negative strip patterns are used. An improved zero-crossing feature detector that has subpixel accuracy is proposed for strip-edge localization. The experimental setup is configured with merely an off-the-shelf pico-projector and a camera. Extensive experiments including accuracy evaluation, comparison with previous structured light algorithms, and the reconstruction of some real shiny objects are shown to demonstrate the systems accuracy and endurance against reflective nature of surfaces.

Laplacian Auto-Encoders

A key factor contributing to the success of many auto-encoders based deep learning techniques is the implicit consideration of the underlying data manifold in their training criteria. In this paper, we aim to make this consideration more explicit by training auto-encoders completely from the manifold learning perspective. We propose a novel unsupervised manifold learning method termed Laplacian Auto-Encoders (LAEs). Starting from a general regularized function learning framework, LAE regularizes training of auto-encoders so that the learned encoding function has the locality-preserving property for data points on the manifold. By exploiting the analog relation between the graph Laplacian and the Laplace-Beltrami operator on the continuous manifold, we derive discrete approximations of the first- and higher-order auto-encoder regularizers that can be applied in practical scenarios, where only data points sampled from the distribution on the manifold are available. Our proposed LAE has potentially better generalization capability, due to its explicit respect of the underlying data manifold. Extensive experiments on benchmark visual classification datasets show that LAE consistently outperforms alternative auto-encoders recently proposed in deep learning literature, especially when training samples are relatively scarce.

A Novel Nonlinear Regression Approach for Efficient and Accurate Image Matting

Current image matting approaches are often implemented based upon color samples under various local assumptions. In this letter, a novel image matting algorithm is investigated by treating the alpha matting as a regression problem. Specifically, we learn spatially-varying relations between pixel features and alpha values using support vector regression. Via the learning-based approach, limitations caused by local image assumptions can be greatly relieved. In addition, the computed confidence terms in learning phase can be conveniently integrated with other matting approaches for the matting accuracy improvement. Qualitative and quantitative evaluations are implemented with a public matting benchmark. And the results are compared with some recent matting algorithms to show its advantages in both efficiency and accuracy.

An incremental PCA-HOG descriptor for robust visual hand tracking

Hand tracking in complicate scenarios is a crucial step to any hand gesture recognition systems. In this paper, we present a novel hand tracking algorithm with adaptive hand appearance modeling. In the algorithm, the hand image is first transformed to the grids of Histograms of Oriented Gradients. And then an incremental Principle Component Analysis is implemented. We name this operator an incremental PCA-HOG (IPHOG) descriptor. The exploitation of this descriptor helps the tracker dealing with vast changing of hand appearances as well as clutter background. Moreover, Particle filter method with certain improvements is also introduced to establish a tracking framework. The experimental results are conducted on an indoor scene with clutter and dynamic background. And the results are also compared with some traditional tracking algorithms to show its strong robustness and higher tracking accuracy.

A Tool-Free Calibration Method for Turntable-Based 3D Scanning Systems

Turntable-based 3D scanners are popular but require calibration of the turntable axis. Existing methods for turntable calibration typically make use of specially designed tools, such as a chessboard or criterion sphere, which users must manually install and dismount. In this article, the authors propose an automatic method to calibrate the turntable axis without any calibration tools. Given a scan sequence of the input object, they first recover the initial rotation axis from an automatic registration step. Then they apply an iterative procedure to obtain the optimized turntable axis. This iterative procedure alternates between two steps: refining the initial pose of the input scans and approximating the rotation matrix. The performance of the proposed method was evaluated on a structured light-based scanning system.

A novel photometric method for real-time 3D reconstruction of fingerprint

3D fingerprint recognition is an emerging technology in biometrics. However, current 3D fingerprint acquisition systems are usually with complex structure and high-cost and that has become the main obstacle for its popularization. In this work, we present a novel photometric method and an experimental setup for real-time 3D fingerprint reconstruction. The proposed system consists of seven LED lights that mounted around one camera. In the surface reflectance modeling of finger surface, a simplified Hanrahan-Krueger model is introduced. And a neural network approach is used to solve the model for accurate estimation of surface normals. A calibration method is also proposed to determine the lighting directions as well as the correction of the lighting fields. Moreover, to stand out the fingerprint ridge features and get better visual effects, a linear transformation is applied to the recovered normals field. Experiments on live fingerprint and the comparison with traditional photometric stereo algorithm are used to demonstrate its high performance.