Jen-Hui Chuang

Regularized Background Adaptation: A Novel Learning Rate Control Scheme for Gaussian Mixture Modeling

To model a scene for background subtraction, Gaussian mixture modeling (GMM) is a popular choice for its capability of adaptation to background variations. However, GMM often suffers from a tradeoff between robustness to background changes and sensitivity to foreground abnormalities and is inefficient in managing the tradeoff for various surveillance scenarios. By reviewing the formulations of GMM, we identify that such a tradeoff can be easily controlled by adaptive adjustments of the GMMs learning rates for image pixels at different locations and of distinct properties. A new rate control scheme based on high-level feedback is then developed to provide better regularization of background adaptation for GMM and to help resolving the tradeoff. Additionally, to handle lighting variations that change too fast to be caught by GMM, a heuristic rooting in frame difference is proposed to assist the proposed rate control scheme for reducing false foreground alarms. Experiments show the proposed learning rate control scheme, together with the heuristic for adaptation of over-quick lighting change, gives better performance than conventional GMM approaches.

Learning a Scene Background Model via Classification

Learning to efficiently construct a scene background model is crucial for tracking techniques relying on background subtraction. Our proposed method is motivated by criteria leading to what a general and reasonable background model should be, and realized by a practical classification technique. Specifically, we consider a two-level approximation scheme that elegantly combines the bottom-up and top-down information for deriving a background model in real time. The key idea of our approach is simple but effective: If a classifier can be used to determine which image blocks are part of the background, its outcomes can help to carry out appropriate blockwise updates in learning such a model. The quality of the solution is further improved by global validations of the local updates to maintain the interblock consistency. And a complete background model can then be obtained based on a measurement of model completion. To demonstrate the effectiveness of our method, various experimental results and comparisons are included.

A probabilistic SVM approach for background scene initialization

Visual tracking systems using background subtraction have been very popular largely due to their efficiency in extracting moving objects. However, such systems often compute the reference background by assuming no moving objects are present during the initialization stage, though the assumption may not be realistic. We propose an automatic way to perform background initialization using a probabilistic SVM (support vector machine). By formulating the problem as an on-line classification one, our approach has the potential to be real-time. SVM classification is carried out for all elements of each image frame by computing the output probabilities. Newly found background elements are evaluated and determined if they should be added to the solution. The process of background initialization continues until there are no more new background elements to be considered. As the features used in an SVM dictate the outcome of classification, we find that optical flow value and inter-frame difference are the two most important ones. Experimental results are included to demonstrate the efficiency of our method.

Vanishing Point-Based Image Transforms for Enhancement of Probabilistic Occupancy Map-Based People Localization

The widespread use of vision-based surveillance systems has inspired many research efforts on people localization. In this paper, a series of novel image transforms based on the vanishing point of vertical lines is proposed for enhancement of the probabilistic occupancy map (POM)-based people localization scheme. Utilizing the characteristic that the extensions of vertical lines intersect at a vanishing point, the proposed transforms, based on image or ground plane coordinate system, aims at producing transformed images wherein each standing/walking person will have an upright appearance. Thus, the degradation in localization accuracy due to the deviation of camera configuration constraint specified can be alleviated, while the computation efficiency resulted from the applicability of integral image can be retained. Experimental results show that significant improvement in POM-based people localization for more general camera configurations can indeed be achieved with the proposed image transforms.

Robust license plate detection in nighttime scenes using multiple intensity IR-illuminator

The functionality of video surveillance is significantly degraded by the low illumination and poor visibility under the nighttime environment. However, the demand for nighttime surveillance is no less than the daytime one because of the high incidence of accidents during night. The Infrared (IR) light source with fixed intensity works for only certain distance, resulting in the defect of underexposure/overexposure due to the object being too far from/close to the light source. In this paper an innovative idea is brought up that we use a multiple intensity IR-illuminator to enhance the effective distance of license plate detection. Based on the stroke width of the license ID, license plates are detected in the images under different illuminations and then the results are integrated into a synthesized high dynamic range image, in which the license plate regions and the background scene can be better visualized. Experimental results show that the proposed approach can effectively enlarge the monitored area in both depth and width, as well as enhance the security level of nighttime video surveillance.

Potential-based modeling of 2-D regions using nonuniform source distributions

One of the existing approaches to path planning problems uses a potential function to represent the topological structure of the free space. Newtonian potential was used in Chuang and Ahuja (1998) to represent object and obstacles in the two-dimensional (2-D) workspace wherein their boundaries are assumed to be uniformly charged. In this paper, more general, nonuniform distributions are considered. It is shown that for linear or quadratic source distributions, the repulsion between two polygonal objects can be evaluated analytically. Simulation results show that by properly adjusting the charge distribution along obstacle/object boundaries, path planning results ran be improved in terms of collision avoidance, path length, etc.

Direct energy minimization for super-resolution on nonlinear manifolds

We address the problem of single image super-resolution by exploring the manifold properties. Given a set of low resolution image patches and their corresponding high resolution patches, we assume they respectively reside on two non-linear manifolds that have similar locally-linear structure. This manifold correlation can be realized by a three-layer Markov network that connects performing super-resolution with energy minimization. The main advantage of our approach is that by working directly with the network model, there is no need to actually construct the mappings for the underlying manifolds. To achieve such efficiency, we establish an energy minimization model for the network that directly accounts for the expected property entailed by the manifold assumption. The resulting energy function has two nice properties for super-resolution. First, the function is convex so that the optimization can be efficiently done. Second, it can be shown to be an upper bound of the reconstruction error by our algorithm. Thus, minimizing the energy function automatically guarantees a lower reconstruction error— an important characteristic for promising stable super-resolution results.

A novel video summarization method for multi-intensity illuminated infrared videos

In nighttime video surveillance, proper illumination plays a key role for the image quality. For ordinary IR-illuminators with fixed intensity, faraway objects are often hard to identify due to insufficient illumination while nearby objects may suffer from over-exposure, resulting in image foreground/background of poor quality. In this paper we proposed a novel video summarization method which utilizes a novel multi-intensity IR-illuminator to generate images of human activities with different illumination levels. By adopting GMM-based foreground extraction procedure for images acquired for each illumination level, foreground objects with most plausible quality can be selected and merged with a preselected representation for still background. The result brings out a reasonable video summary for moving foreground, which is generally unachievable for nighttime surveillance videos.

Vanishing Point-based Line Sampling for Real-time People Localization

In this paper, we propose a real-time multicamera people localization method based on line sampling of image foregrounds. For each view, these line samples are originated from the vanishing point of lines perpendicular to the ground plane. With these line samples, vertical line samples in the 3-D scene can be reconstructed for potential human locations. After some efficient geometric refinement and filtering procedures, the remaining qualified 3-D line samples are clustered and integrated for the identification of locations and heights of people in the scene. Both indoor and outdoor scenarios are examined to demonstrate the effectiveness of our approach in handling serious occlusion in crowed scenes. The average localization error of less than 15 cm for average viewing distance of 15m suggests that our method can be applied to a broad range of surveillance applications that require the real-time computation of localization without using special hardware for acceleration.

Vanishing point-based line sampling for efficient axis-based people localization

In this paper, we propose an efficient people localization approach using multiple cameras based on axial representations of foreground regions. Unlike many previous methods that need to project all foreground pixels of all views to multiple reference planes via homography, we instead apply vanishing point-based line sampling to reduce the large amount of pixel processing so that computational efficiency can be greatly enhanced. Experimental simulations show that the proposed approach is more than 40 times faster than the compared, pixel-based localization method on average, without sacrificing the localization accuracy.