Changjiang Yang

Efficient mean-shift tracking via a new similarity measure

The mean shift algorithm has achieved considerable success in object tracking due to its simplicity and robustness. It finds local minima of a similarity measure between the color histograms or kernel density estimates of the model and target image. The most typically used similarity measures are the Bhattacharyya coefficient or the Kullback-Leibler divergence. In practice, these approaches face three difficulties. First, the spatial information of the target is lost when the color histogram is employed, which precludes the application of more elaborate motion models. Second, the classical similarity measures are not very discriminative. Third, the sample-based classical similarity measures require a calculation that is quadratic in the number of samples, making real-time performance difficult. To deal with these difficulties we propose a new, simple-to-compute and more discriminative similarity measure in spatial-feature spaces. The new similarity measure allows the mean shift algorithm to track more general motion models in an integrated way. To reduce the complexity of the computation to linear order we employ the recently proposed improved fast Gauss transform. This leads to a very efficient and robust nonparametric spatial-feature tracking algorithm. The algorithm is tested on several image sequences and shown to achieve robust and reliable frame-rate tracking.

Fast multiple object tracking via a hierarchical particle filter

A very efficient and robust visual object tracking algorithm based on the particle filter is presented. The method characterizes the tracked objects using color and edge orientation histogram features. While the use of more features and samples can improve the robustness, the computational load required by the particle filter increases. To accelerate the algorithm while retaining robustness we adopt several enhancements in the algorithm. The first is the use of integral images for efficiently computing the color features and edge orientation histograms, which allows a large amount of particles and a better description of the targets. Next, the observation likelihood based on multiple features is computed in a coarse-to-fine manner, which allows the computation to quickly focus on the more promising regions. Quasi-random sampling of the particles allows the filter to achieve a higher convergence rate. The resulting tracking algorithm maintains multiple hypotheses and offers robustness against clutter or short period occlusions. Experimental results demonstrate the efficiency and effectiveness of the algorithm for single and multiple object tracking.

Automatic image orientation detection

We present an algorithm for automatic image orientation estimation using a Bayesian learning framework. We demonstrate that a small codebook (the optimal size of codebook is selected using a modified MDL criterion) extracted from a learning vector quantizer (LVQ) can be used to estimate the class-conditional densities of the observed features needed for the Bayesian methodology. We further show how principal component analysis (PCA) and linear discriminant analysis (LDA) can be used as a feature extraction mechanism to remove redundancies in the high-dimensional feature vectors used for classification. The proposed method is compared with four different commonly used classifiers, namely k-nearest neighbor, support vector machine (SVM), a mixture of Gaussians, and hierarchical discriminating regression (HDR) tree. Experiments on a database of 16 344 images have shown that our proposed algorithm achieves an accuracy of approximately 98% on the training set and over 97% on an independent test set. A slight improvement in classification accuracy is achieved by employing classifier combination techniques.

Mean-shift analysis using quasiNewton methods

Mean-shift analysis is a general nonparametric clustering technique based on density estimation for the analysis of complex feature spaces. The algorithm consists of a simple iterative procedure that shifts each of the feature points to the nearest stationary point along the gradient directions of the estimated density function. It has been successfully applied to many applications such as segmentation and tracking. However, despite its promising performance, there are applications for which the algorithm converges too slowly to be practical. We propose and implement an improved version of the mean-shift algorithm using quasiNewton methods to achieve higher convergence rates. Another benefit of our algorithm is its ability to achieve clustering even for very complex and irregular feature-space topography. Experimental results demonstrate the efficiency and effectiveness of our algorithm.

Planar conic based camera calibration

Inspired by the technique proposed by Zhang (1998), we proposed a camera calibration technique, which only requires observing three or more planar concentric conics at a few (at least two) different orientations. All computations involved are linear matrix manipulations. Compared with the classical techniques where an expensive calibration pattern is commonly used, our technique is easy to implement and more flexible. Using conics also simplifies the problem of correspondence. Both computer simulation and real data are used to test the proposed technique.

Super-resolution using preconditioned conjugate gradient method

In this paper we present a fast iterative image superresolution algorithm using preconditioned conjugate gradient method. To avoid explicitly computing the tolerance in the inverse filter based preconditioner scheme,1 a new Wiener filter based preconditioner for the conjugate gradient method is proposed to speed up the convergence. The circulant-block structure of the preconditioner allows efficient implementation using Fast Fourier Transform. Effectiveness of the preconditioner is demonstrated by superresolution results for simulated image sequences.

An intrinsic parameters self-calibration technique for active vision system

This paper presents a new camera intrinsic parameters self-calibration technique for an ordinary active vision system. By controlling a pan-tilt-translation camera platform to do a sequence of specially designed motions, called a camera motion configuration, we rigorously proved that the camera intrinsic parameters can be determined linearly under such two configurations: 1) regulating the cameras orientation by 3 tilts, at each cameras orientation, controlling the camera to translate twice along 2 orthogonal directions; 2) regulating cameras orientation by 1 pan and 2 tilts, at each cameras orientation, controlling the camera to translate twice along 2 orthogonal directions. It is shown that the configuration 2 is robust, whereas the configuration 1 is numerically unstable and sensitive to noise. Experiments with real data were carried out and the calibration results have been verified by a stereo vision experiment. A comparison with other camera calibration approaches is also reported.

Near-optimal regularization parameters for applications in computer vision

Computer vision requires the solution of many ill-posed problems such as optical flow, structure from motion, shape from shading, surface reconstruction, image restoration and edge detection. Regularization is a popular method to solve ill-posed problems, in which the solution is sought by minimization of a sum of two weighted terms, one measuring the error arising from the ill-posed model, the other indicating the distance between the solution and some class of solutions chosen on the basis of prior knowledge (smoothness, or other prior information). One of important issues in regularization is choosing optimal weight (or regularization parameter). Existing methods for choosing regularization parameters either require the prior information on noise in the data, or are heuristic graphical methods. We apply a method for choosing near-optimal regularization parameters by approximately minimizing the distance between the true solution and the family of regularized solutions. We demonstrate the effectiveness of this approach for the regularization on two examples: edge detection and image restoration.

Visual motion based behavior learning using hierarchical discriminant regression

This paper presents a new technique which incrementally builds a hierarchical discriminant regression (IHDR) tree for generation of motion based robot reactions. The robot learned the desired reactions from motion change images, without using other pre-defined features. The generation from training cases is accomplished through the automatically constructed IHDR tree, which automatically derives features that are most related to outputs and disregards subspaces that are not related, or less related, to outputs. The real-time speed is achieved through combination of feature space partition and a coarse-to-fine sample search, which results in a logarithmic time complexity in the number of nodes. The experiments showed that the IHDR method can interpolate the mapping between high dimensional input space and the output control signal space from a variety of objects of various shapes with different motion patterns, based on the size-dependent negative logarithmic distance measures in the hierarchical feature space. The trained robot is able to aim to its camera towards moving object and move toward or away according to the size of moving object.

An inherent probabilistic aspect of the Hough transform

In this paper, a new property of the Hough transform is discovered, namely an inherent probabilistic aspect which is independent of the input image and embedded in the transformation process from the image space to the parameter space. It is shown that such a probabilistic aspect has a wide range of implications concerning the specification of implementation schemes and the performance of Hough transform. In particular, it is shown that in order to make the Hough transform really meaningful, an appropriate curve (surface) density function must be, either explicitly or implicitly, supplied during its implementation process, and that the widely used approach to uniformly discretizing parameter space in the literature is generally inadequate.