Yang Weon Lee

Adaptive data association for multi-target tracking using relaxation

This paper introduces an adaptive algorithm determining the measurement-track association problem in multi-target tracking. We model the target and measurement relationships and then define a MAP estimate for the optimal association. Based on this model, we introduce an energy function defined over the measurement space, that incorporates the natural constraints for target tracking. To find the minimizer of the energy function, we derived a new adaptive algorithm by introducing the Lagrange multipliers and local dual theory. Through the experiments, we show that this algorithm is stable and works well in general environments.

Application of the Particle Filter for Simple Gesture Recognition

This paper describes a gesture recognition algorithm based on the particle filters, namely CONDENSATION. The particle filter is more efficient than any other tracking algorithm because the tracking mechanism follows Bayesian estimation rule of conditional probability propagation.

Implementation of Mutual Localization of Multi-robot Using Particle Filter

This paper describes an implementation of mutual localization of swarm robot using particle filter. Robots determine the location of the other robots using wireless sensors. Measured data will be used for determination of the robot itself moving method. It also effects on the other robot’s formation such as circle and line type formation. We discuss the problem in circle formation enclosing target which moves. This method is the solution about enclosed invader in circle formation based on mutual localization of multi-robot without infrastructure. We use trilateration which does not need to know the value of the coordinates of reference points. So, specify enclosed point for the number of robots based on between the relative positions of the robot in the coordinate system. Particle filter is used to improve the accuracy of the robot’s location. The particle filter is well operated for mutual location of robots than any other estimation algorithm. Through the experiments, we show that the proposed scheme is stable and works well in real environments

Implementation of interactive interview system using hand gesture recognition

This paper describes a implementation of virtual interactive interview system. A hand motion recognition algorithm based on the particle filters is applied for this system. The particle filter is well operated for human hand motion recognition than any other recognition algorithm. Through the experiments, we show that the proposed scheme is stable and works well in virtual interview systems environments.

Context-Based approach for human gesture analysis

In this paper, we propose a state transition model using context based approach for gesture analysis. This method defines the analysis situation as five different states; NULL, OBJECT, POSTURE, LOCAL, and GLOBAL. We first infer the situation of the system by estimating the transition of the state model, and then apply different analysis algorithms according to the system state. Gestures are analyzed with the queue-based matching method which is newly proposed in the paper instead of general gesture spotting algorithms. In the algorithm, movement of feature points; face and both hands, is compared directly, so gesture can be recognized quickly without applying any constraints for real world application. The transition of the states can be interpreted as the context of motion and that is estimated with conditional probability between the states in the algorithm.

A study on the correction of gun fire error using neural network

Multilayer perceptrons trained with the backpropagation algorithm are derived for gun fire control system for miss distance correction and are compared to optimum linear filter based on minimum mean square error [1], [2]. The structure of the proposed neural controller is described and performance results are shown.

Development of Automatic Code Generation Tool for Condensation Algorithm

This paper address the problem of tracking multiple objects encountered in many situations in developing condensation algorithms. The difficulty lies on the fact that the implementation of condensation algorithm is not easy for the general users. We propose an automatic code generation program for condensation algorithm using MATLAB tool. It will help for general user who is not familiar with condensation algorithm to apply easily for real system. The merit of this program is that a general industrial engineer can easily simulate the designed system and confirm the its performance on the fly.

A study on the gesture recognition based on the particle filter

The recognition of human gestures in image sequences is an important and challenging problem that enables a host of human-computer interaction applications. This paper describes a gesture recognition algorithm based on the particle filters, namely CONDENSATION. The particle filter is more efficient than any other tracking algorithm because the tracking mechanism follows Bayesian estimation rule of conditional probability propagation. We used two models for the evaluation of particle filter and apply the MAILAB for the preprocessing of the image sequence. But we implement the particle filter using the C++ to get the high speed processing. In the experimental results, it is demonstrated that the proposed algorithm prove to be robust in the cluttered environment.

Development of the Practical Particle Filter for Human Gesture Recognition

The recognition of human gestures in image sequences is an important and challenging problem that enables a host of human-computer interaction applications. This paper describes a gesture recognition algorithm based on the particle filters, namely CONDENSATION. The particle filter is more efficient than any other tracking algorithm because the tracking mechanism follows Bayesian estimation rule of conditional probability propagation. We used two models for the evaluation of particle filter and apply the MATLAB for the preprocessing of the image sequence. But we implement the particle filter using the C++ to get the high speed processing. In the experimental results, it is demonstrated that the proposed algorithm prove to be robust in the cluttered environment.

Design of data association filter using neural networks for multi-target tracking

In this paper, we have developed the MHDA scheme for data association. This scheme is important in providing a computationally feasible alternative to complete enumeration of JPDA which is intractable. We have proved that given an artificial measurement and track’s configuration, MHDA scheme converges to a proper plot in a finite number of iterations. Also, a proper plot which is not the global solution can be corrected by re-initializing one or more times. In this light, even if the performance is enhanced by using the MHDA, we also note that the difficulty in tuning the parameters of the MHDA is critical aspect of this scheme. The difficulty can, however, be overcome by developing suitable automatic instruments that will iteratively verify convergence as the network parameters vary.