This paper introduces an adaptive visual tracking method that combines the adaptive appearance model and the optimization capability of the Markov decision process. Most tracking algorithms are limited due to variations in object appearance from changes in illumination, viewing angle, object scale, and object shape. This paper is motivated by the fact that tracking performance degradation is caused not only by changes in object appearance but also by the inflexible controls of tracker parameters. To the best of our knowledge, optimization of tracker parameters has not been thoroughly investigated, even though it critically influences tracking performance. The challenge is to equip an adaptive tracking algorithm with an optimization capability for a more flexible and robust appearance model. In this paper, the Markov decision process, which has been applied successfully in many dynamic systems, is employed to optimize an adaptive appearance model-based tracking algorithm. The adaptive visual tracking is formulated as a Markov decision process based dynamic parameter optimization problem with uncertain and incomplete information. The high computation requirements of the Markov decision process formulation are solved by the proposed prioritized Q-learning approach. We carried out extensive experiments using realistic video sets, and achieved very encouraging and competitive results.

This paper presents an evolutionary and adaptive framework for efficient visual tracking based on a hybrid POMDP formulation. The main focus is to guarantee visual tracking performance under varying environments without strongly-controlled situations or high-cost devices. The performance optimization is formulated as a dynamic adaptation of the system control parameters, i.e., threshold and adjusting parameters in a visual tracking algorithm, based on the hybrid of offline and online POMDPs. The hybrid POMDP allows the agent to construct world-belief models under uncertain environments in offline, and focus on the optimization of the system control parameters over the current world model in real-time. Since the visual tracking should satisfy strict real-time constraints, we restrict our attention to simpler and faster approaches instead of exploring the belief space of each world model directly. The hybrid POMDP is thus solved by an evolutionary adaptive framework employing the GA (Genetic Algorithm) and real-time Q-learning approaches in the optimally reachable genotype and phenotype spaces, respectively. Experiments were carried out extensively in the area of eye tracking using videos of various structures and qualities, and yielded very encouraging results. The framework can achieve an optimal performance by balancing the tracking accuracy and real-time constraints.

Effective perimeter protection mechanisms for industrial sites and critical infrastructure must contend with a large variety of potential threats as well as with the fact that normal site activity can be both complex and diverse. This paper documents the development of a system level approach capable of functioning under such challenging conditions. A multi-view tracking system is used to provide real-time site wide trajectories of all observed individuals. A Radar-based system is also used for tracking if and when camera coverage of various regions is not available. Track information is then analyzed with respect to articulated motion analysis, complex event analysis and normalcy analysis. In addition, object recognition is used to classify left behind objects using high resolution PTZ imagery. A real-time integrated version of this comprehensive approach to perimeter protection was deployed using a single standard off-the-shelf desktop computer.

Summary form only given. The rapid evolution of tools and software systems to design experiments, automatically monitor, collect and warehouse large amounts of data, from applications such as life sciences and industrial processes has resulted in a new paradigm shift. This change of paradigm is so fast that some of the practices for optimization and management of these processes that were valid only 5–10 years ago may no longer be fully acceptable or sufficient for today's business optimization and management. This has a direct influence on the best practices for knowledge discovery and management of the discovered knowledge in real-world data mining applications. Establishing and managing a real-world data mining project in any domain, in particular in today's life science industry, is not a trivial task. A few approaches have been proposed in the literature. However, initiation and successful management of such efforts may depend on where a given case study fits in the overall classification of data mining approaches. Today's knowledge discovery from data can be classified in several ways: (i) data mining on engineered systems (e.g. complex equipment) or systems designed by nature (e.g. life sciences), (ii) explanatory or predictive data mining, (iii) data mining from static data (e.g. data warehouse) or dynamic data (e.g. data streams), (iv) user operated or automated data mining. There could still be other ways to classify data mining applications. This talk provides an overview of the above listed knowledge discovery applications. We provide examples where we demonstrate how small or large amounts of data, when understood from a real-world data mining point of view and the required data is properly integrated, can result in novel knowledge discovery case studies. We explain motivations and challenges of establishing real-world dat

In this paper, we proposed a new facial landmark-detection system using as edge energy function. The facial landmark-detection system is divided into a learning stage and a detection stage. The learning stage creates an interest-region model, to set up a search region of each landmark, as pre-information necessary for a detection stage and creates a detector for each landmark to detect a landmark in a search region. The detection stage sets up a search region of each landmark in an input image with an interest-region model created in the learning stage. Because a landmark to detect from a system has the characteristics of an edge as both edge of an eye, both edge of a mouth and both edges of eyebrows, we have detected a landmark by applying an edge energy function to the Bayesian discrimination method. We have implemented aforementioned technique by abstracting 800 impassive images from the FERET database and have measured data in which the normalized average error distance is less than 0.1 occupying 98% of the total data.

In this paper, we propose an online evolutionary context-aware classifier ensemble framework for object recognition systems which are adaptive to various environments. The starting point utilizes a context recognizer, a context knowledge base and a classifier ensemble generator to provide optimal solutions for classifier ensemble. Even though our framework is quite general and could be applied to various classification tasks, here we focus on the cases of face recognition. The proposed framework uses an unsupervised learning method to carry out context modeling tasks for various environments. The data for classifier ensemble are assigned to corresponding contexts based on supervised learning and an evolutionary algorithm processes all the information to generate context knowledge for online adaptation. Experimental comparisons with systems based on conventional face recognition algorithms upon four extended benchmark data sets, E-FERET, E-Yale, E-INHA, and our own database showed that the system based on our framework was able to operate in dynamic environments with stable performance which others could not achieve.

We propose the use of the personalized ontology model to improve the effectiveness of web documents filtering process. One important feature of this model is that by constructing the user specific ontology, web documents can be classified by using the user oriented meta data that reflects the user's view about the documents concept. Another is that by applying the user model to searching the classified documents, we achieved the effective document search performance. To find the user's preference, Bayesian Learner accepts user's interests flow as an input and writes output to users profile. Based on those user profiles, user specific ontologies are constructed to provide efficient search environment.

Face image processing techniques are essential for advanced human-computer interaction. Most previous face image processing systems have been proposed under the assumption that approximative face regions or the locations of facial features are known. In this paper, we mainly discuss face and eye location algorithms. The proposed techniques use no prior knowledge on the given image. It is less sensitive to backgrounds and lighting conditions than previous approaches. Mosaic image representation and valley representation are adopted for face detection and eye location, respectively. Extensive experiments have been performed, and very encouraging results have been achieved.

In this paper, we present the framework design of IDAP (Intelligent Document Assistant Processor) with the capability of intelligent user interface for pen-based computer systems. No conventional computer user interface can compete with pen and paper until now in terms of convenience and ease of use in generating documents. The system gets roughly drawn documents from users, and generates neat documents satisfiable to users. Rough document includes handwritten texts, hand-drawn tables, hand-drawn diagrams, handwritten mathematical expressions. The satisfaction of users is at least as important as the functionality and performance in pen-based system. Therefore, sophisticated user interface is critical for a successful design. Intelligent user interface methodology is employed to make the system more intelligent, natural, and ease of use. Human-computer interaction model called H-COS is proposed for pen-based applications, and the model is applied to design IDAP.

This paper addresses a novel recognition technique for rotated character strings. Rotated character strings appear in character/graph mixed drawings such as cadastral maps, engineering drawings, etc. Recently, there are huge demands for systems of automatic map recognition. One of the most critical obstacles for such systems, however is the recognition of rotated character strings. Rotated characters are represented by the relative chain code and recognized. The possible misinterpretation of a character string has been resolved by deriving vertical/horizontal vector of the character string from a recognized unambiguous character in the character string. The proposed technique discusses primarily numeral string recognition, but it can be readily extended to the strings of other character sets.

The authors propose to set up assignment rules based on the available interconnection resources of 2-D processor arrays. The reconfiguration of 2-D processor arrays is guided by the assignment rules, such that logical cells are always connected through the available interconnection resources. The advantage of the proposed approach is the easy adaptation to various system requirements, by adjusting assignment rules. The proposed reconfiguration algorithm can be extended to arrays with clustered faulty cells, overcoming the weakness of fixed domain approaches.>

A novel reconfiguration scheme is proposed to achieve a highly successful reconfiguration rate for two-dimensional processor arrays with fixed interconnection resources. The authors use a flexible spare allocation scheme and thereby achieve a spare allocation pattern which maximizes the utilization of the given interconnection resources. However, some spare allocations may not result in the successful reconfiguration of a logic array due to the lack of necessary resources. The rationale behind this approach is that the probability of the occurrence of allocation patterns causing such resource conflicts is pretty low. Furthermore, the rare resource conflicts may be resolved by the proposed resolution techniques. By utilizing the given interconnection resources much more efficiently, the proposed scheme achieves a higher reconfiguration rate than a previous design of similar approach. The complexity of the reconfiguration algorithm is O(N/sup 2/).>