• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.23 no.9A
• /
• pp.2355-2364
• /
• 1998

In conventional fractal decoding procedure, the reconstructed image is obtained by a rpredefined number of iterations starting with an arbitrary initial image. Its convergence speed depends on the selection of the initial image. It should be solved to get high speed convergence. In this paper, we theoretically reveal that conventional method is approximately decomposed into the decoding of the DC and AC components. Based on this fact, we proposed a novel fast fractal decoding algorithm made up of two steps. The averaged-image considered as an optimal initial image is estimated in the first step. In the second step, the reconstructe dimag eis genrated from the output image obtained in the first step. From the simulations, it is shown that the output image of the first step approximately converges to the averaged-image with only 15% calculations for one iteration of conventional method. And the proposed method is faster than various decoding mehtods and evenly equal to conventioanl decoding with the averaged-image. In addition, the proposed method can be applied to the compressed data resulted from the various encoding methods because it does not impose any constraints in the encoding procedure to get high decoding speed.

• Proceedings of the IEEK Conference
• /
• 2000.07a
• /
• pp.326-328
• /
• 2000

The paper proposes the scheme of skew detection for Thai printed document images by using linear regression algorithm. It intends to use with the Thai character recognition systems to reduce the skew detection time. This scheme begins by finding the center of gravity of a document image. This point is used as the starting point for gathering data in the scheme. The data is obtained by scanning incrementally one pixel in vertically with the width of 20-pixels. After the scanning process, if data Is different from it's neighbor more than ${\pm}$ 15 pixels, it will be considered as noise or data in other lines and will be deleted. The last step is the operation by using linear regression algorithm on these selected data and the skew angle will be obtained. The proposed method has been tested with 45 document images with different fonts, sizes and skew angles. The experiment results show that the proposed method can detect the skew angle with the error of less then one degree. The average processing time is about 19 times faster than that of the Hough Transform method.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• v.1
• /
• pp.259-264
• /
• 2006

It is well known that the uncertainty of the covariance parameters of the process noise (Q) and the observation errors (R) has a significant impact on Kalman filtering performance. Q and R influence the weight that the filter applies between the existing process information and the latest measurements. Errors in any of them may result in the filter being suboptimal or even cause it to diverge. The conventional way of determining Q and R requires good a priori knowledge of the process noises and measurement errors, which normally comes from intensive empirical analysis. Many adaptive methods have been developed to overcome the conventional Kalman filter's limitations. Starting from covariance matching principles, an innovative adaptive process noise scaling algorithm has been proposed in this paper. Without artificial or empirical parameters to be set, the proposed adaptive mechanism drives the filter autonomously to the optimal mode. The proposed algorithm has been tested using road test data, showing significant improvements to filtering performance.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.10 no.4
• /
• pp.373-379
• /
• 2005

In this paper, the control algorithm of DC source device for inverter starting is proposed and the control method for compensating unbalance system source on operating time in the voltage type PWM converter with driving and regenerative faculty is suggested. The maintaining way of balancing condition for converter of AC source is used the compensating unbalanced status by current control loop. Because it is possible that the unbalanced System control is used to leakage transformer not equaled reactance by each phase in rectifier system, the proposed H/W and control algorithm of rectifier system is contributed to minimize of device and rising efficiency.

• Journal of KIISE:Software and Applications
• /
• v.27 no.2
• /
• pp.97-107
• /
• 2000

Incremental learning generally reduces training time and increases the generalization of a neural network by selecting training data incrementally during the training. However, the existing methods of incremental learning repeatedly evaluate the importance of training data every time they select additional data. In this paper, an incremental learning algorithm is proposed for pattern classification problems. It evaluates the importance of each piece of data only once before starting the training. The importance of the data depends on how close they are to the decision boundary. The current paper presents an algorithm which orders the data according to their distance to the decision boundary by using clustering. Experimental results of two artificial and real world classification problems show that this proposed incremental learning method significantly reduces the size of the training set without decreasing generalization performance.

• The Journal of Korea Robotics Society
• /
• v.5 no.2
• /
• pp.110-119
• /
• 2010

This paper aims to add the autonomous driving capability to the inverted pendulum system which maintains the inverted pendulum upright stably. For the autonomous driving from the starting position to the goal position, the motion control algorithm is proposed based on the dynamics of the inverted pendulum robot. To derive the dynamic model of the inverted pendulum robot, a three dimensional robot coordinate is defined and the velocity jacobian is newly derived. With the analysis of the wheel rolling motion, the dynamics of inverted pendulum robot are derived and used for the motion control algorithm. To maintain the balance of the inverted pendulum, the autonomous driving strategy is derived step by step considering the acceleration, constant velocity and deceleration states simultaneously. The driving experiments of inverted pendulum robot are performed while maintaining the balance of the inverted pendulum. For reading the positions of the inverted pendulum and wheels, only the encoders are utilized to make the system cheap and reliable. Even though the derived dynamics works for the slanted surface, the experiments are carried out in the standardized flat ground using the inverted pendulum robot in this paper. The experimental data for the wheel rolling and inverted pendulum motions are demonstrated for the straight line motion from a start position to the goal position.

• Industrial Engineering and Management Systems
• /
• v.9 no.2
• /
• pp.131-140
• /
• 2010

Biclustering is a method of finding meaningful subsets of objects and attributes simultaneously, which may not be detected by traditional clustering methods. It is popularly used for the analysis of microarray data representing the expression levels of genes by conditions. Usually, biclustering algorithms do not consider a sequential relation between attributes. For time series data, however, bicluster solutions should keep the time sequence. This paper proposes a new biclustering algorithm for time series data by modifying the plaid model. The proposed algorithm introduces a parameter controlling an interval between two selected time points. Also, the pruning step preventing an over-fitting problem is modified so as to eliminate only starting or ending points. Results from artificial data sets show that the proposed method is more suitable for the extraction of biclusters from time series data sets. Moreover, by using the proposed method, we find some interesting observations from real-world time-course microarray data sets and apartment price data sets in metropolitan areas.

• MALSORI
• /
• no.58
• /
• pp.67-81
• /
• 2006

For a large corpus of time-aligned data, HMM based approaches are most widely used for automatic segmentation, providing a consistent and accurate phone labeling scheme. There are two methods for training in HMM. Flat starting method has a property that human interference is minimized but it has low accuracy. Bootstrap method has a high accuracy, but it has a defect that manual segmentation is required In this paper, a new algorithm is proposed to minimize manual work and to improve the performance of automatic segmentation. At first phase, voiced, unvoiced and silence classification is performed for each speech data frame. At second phase, the phoneme sequence is aligned dynamically to the voiced/unvoiced/silence sequence according to the acoustic phonetic rules. Finally, using these segmented speech data as a bootstrap, phoneme model parameters based on HMM are trained. For the performance test, hand labeled ETRI speech DB was used. The experiment results showed that our algorithm achieved 10% improvement of segmentation accuracy within 20 ms tolerable error range. Especially for the unvoiced consonants, it showed 30% improvement.

• Journal of the military operations research society of Korea
• /
• v.26 no.1
• /
• pp.89-99
• /
• 2000

This paper deals with a heuristic algorithm for the vehicle routing-scheduling problem to minimize the total travel distance and the total cost. Because the aim of the Clarke-Wright method, one of famous heuristic methods, is to minimize the total travel distance of vehicles, it cannot consider the cost if the cost and the travel distance is not proportional. In the Clarke-Wright method, the route of each vehicle is found by using the saving matrix which is made by an assumption that the vehicle comes back to the starting point. The problem dealt with in the paper, however, does not need the vehicle to come back because each vehicle has its hoping-start-points and hoping-destination-points. Therefore we need a different saving matrix appropriate to this occasion. We propose a method to find an initial solution by applying network simplex method after transforming the vehicle routing-scheduling problem into the minimum cost problem. Moreover, we propose a method to minimize the total travel distance by using the modified saving matrix which is appropriate to no-return occasion and the method for the case of plural types of vehicles and freights.

• Journal of the Korean Society for Precision Engineering
• /
• v.17 no.8
• /
• pp.157-164
• /
• 2000

This paper presents a method to detect a rotor position and to drive a BLDC motor from standstill to medium speed without any position sensor comparing the current responses due to the inductance variation in the rotor position. A rotor position at a standstill is identified by the current responses of six pulses injected to each phase of a motor. Once the motor stars up pulse train that is composed of long and short pulses is injected to the phase corresponding to produce the maximum torque and the next phase continuously. it provides not only the torque but also the information of the next commutation time effectively when the response of long and short pulses crosses each other after the same time delay. This method which is verified experimentally using a DSP can drive a BLDC motor to the medium speed smoothly without any rattling and time delay compared with the conventional sensorless algorithm.