• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.12
• /
• pp.1624-1634
• /
• 1995

Conventional full search block matching algorithm for motion estimation is computationally intensive and the resulting hardware cost is very high. In this paper, we present an efficient block matching algorithm using a 16:1 subsampling technique, and describe its hardware design. The algorithm reduces the number of pixels in calculating the mean absolute difference at each search location, instead of reducing the search locations.The algorithm is an extension of the block mating algorithm with 4:1 subsampling proposed by Liu and Zaccarin such that the amount of computation is reduced by a fact of 4(16 compared to the full search block matching algorithm) while producing similar performance.The algorithm can efficiently be designed into a hardware for real-time applications.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.9
• /
• pp.142-149
• /
• 1994

Hangul is expressed by the basic elements, twenty-four characters. Because these characters are composed of a circle and lines, Hough transform(HT), which has a powerful performance on the noise in extracting lines, is introduced. Many difficulties often occur when the original HT is used to extract strokes and it's direction, position and length from handwritten Hangul characters. Original HT has eight direction selected as samples in the transformed image should be calculated for these eight directions. In this paper, the generalized sampling rule is suggested. According to the rule, those directions which are possible to a line are the only thing to be calculated. The experoment result turned out to be higher than the method that Chen suggested in sampling rate. Anogher experiment result is done on the 1800 handwritten Hangul characters that 10 persons wrote. By feature extracting the oritinal HT and sampling HT. And as a result of six type classification, the suggested method came out higher than original HT.

• Proceedings of The KACE
• /
• 2018.01a
• /
• pp.41-44
• /
• 2018

교육데이터마이닝은 다양한 교육 환경에서 생성되는 막대한 양의 데이터를 활용하여 학습자들의 학습 유형, 학습 진도를 분석, 예측하고 교육 성취를 효과적으로 향상시키는 것을 목적으로 한다. 효과적인 교육데이터마이닝 결과를 얻기 위해서는 교육데이터에 대한 정제 과정이 필요하며 DBSCAN 클러스터링을 통해 교육데이터에 포함된 노이즈 데이터를 제거하고 생성된 각 클러스터에서 동일한 비율로 데이터를 추출함으로써 편향되지 않은 표본 데이터를 생성할 수 있다. 하지만 DBSCAN은 두 개의 전역 매개변수에 의해 다양한 밀도분포를 가지는 클러스터를 생성할 수 없다는 문제점이 있으며 이는 교육 데이터를 정제함에 있어 치명적인 문제점이 될 수 있다. 본 논문에서는 DBSCAN의 문제점을 개선하고 클러스터링 정확도를 향상시키기 위해 고정된 매개변수를 사용하지 않고 각 밀도분포에 대해 최적의 입력 매개변수를 결정함으로써 다양한 밀도분포를 가지는 클러스터들을 효과적으로 생성하는 C-DBSCAN을 제안한다.

• Proceedings of the IEEK Conference
• /
• 2006.06a
• /
• pp.269-270
• /
• 2006

내용 기반 영상검색을 위한 효과적 특징 추출 방법에 대한 관심이 집중되고 있다. 본 논문에서는 물체의 크기에 강인한 영상검색을 위한 색상 특징정보와 상관관계를 이용한 RGB 칼라 검색 기법을 제안하였다. R, G, B 영상을 각 칼라 R, G, B 크기 순으로 배열한 후 칼라 R, G, B 간의 분산 차이를 특징 벡터로 사용하였다. 그리고 각 계산된 분산차를 3개의 구역으로 표본화여 재배열하였다. 제안된 알고리즘은 실험을 통하여 기존의 칼라 히스토그램, CCV, 칼라 코렐로그램 보다 검색 성능이 더 향상됨을 보였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.19 no.10
• /
• pp.1991-2000
• /
• 1994

In this paper, an alagorithm with six types of classification is suggested for the recognition system of hand-written Korean characters. After thinning process and truncating process for noise redection. The input images are used generalized by $64\times64$ size. The six type classification is composed of preliminary and secondary classification process by using the learning algoritm of multi-layer perceptron. Subblock Hough transform is used as local feature and sampling Hough transform is used as global feature. Experiment is conducted for 1800 characters which is written 31 times per each type by 10 persons. The 90% recognition rate is resulted by the preliminary classification of detection the final consonant and by the secondary classification of detecting the vowels.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.11A
• /
• pp.1864-1871
• /
• 2001

In this paper, new symbol timing recovery algorithm is proposed, which is suitable for the digital communication system with Multi-level PAM signals. In the newly proposed symbol timing recovery algorithm, the timing error function is derived by compensating the several difference values between sampled symbol and neighboring symbol every symbol period with mid samples and decided symbol values. Conventional symbol timing recovery methods did not work well in Multi-level PAM signals, but the newly proposed method can be applied to Multi-level PAM signals as well as QPSK. For the performance analysis, the derived variance of the timing error function and the timing error characteristics of S-curve show that the proposed method has better performance than Gardner method and the modified Gardner method.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.11 no.4
• /
• pp.99-104
• /
• 2011

In this paper the approach of neural network was proposed which detects a variety of defects in the molded parts. In an attempt to improve the response of the system, It is designed to minimize the use of memory via LookUp table in software. The goal of these methods was to extract the features of samples in learning of neural networks, overcoming the algorithms of defects detection and classification. Through the learning of 500 sample patterns of molded parts, defects of 3% molded parts was detected and classified as the incorrect diameter parts. We expect that proposed approach is an effective alternative to save test time and cost for defect detection of a fine pattern within the molded parts.

• Journal of Cadastre & Land InformatiX
• /
• v.46 no.1
• /
• pp.87-100
• /
• 2016

The acquisition of the forest resource information has depended on a partial sampling method or aerial photographs which demand a lot of effort and time because of the vast areas and the difficult approach. For the acquisition of the forest resource information, there have been the optical remote-sensing and the multi-spectrum image to offer only horizontal distributions of trees, but a new technological approach, such as Airborne LiDAR, is more necessary to acquire directly three dimensional information related to the forest terrains and trees' features. This paper proposes an algorithm for the forest information extraction such as trees' individual numbers and the heights of trees by using LiDAR data. Especially, this proposed algorithm adopts a region growing method for the extraction of the vegetation-point and extracts the forest information using morphological features of trees.

• The Korean Journal of Applied Statistics
• /
• v.33 no.4
• /
• pp.451-466
• /
• 2020

CCC-r chart is effective for high-quality processes with a very low fraction nonconforming. The values of process parameters should be estimated from the Phase I sample since they are often not known. However, if the Phase I sample size is not sufficiently large, an estimation error may occur when the parameter is estimated and the practitioner may not achieve the desired in-control performance. Therefore, we adjust the control limits of CCC-r charts using the bootstrap algorithm to improve the in-control performance of charts with smaller sample sizes. The simulation results show that the adjustment with the bootstrap algorithm improves the in-control performance of CCC-r charts by controlling the probability that the in-control average number of observations to signal (ANOS) has a value greater than the desired one.

• The Journal of Natural Sciences
• /
• v.15 no.1
• /
• pp.11-34
• /
• 2005

It is no exaggeration to say that the productivity of a research using computer simulations on complex molecular systems like biomolecules depends on the ability of the sampling algorithm to explore the relevant parts of configuration space. In this study, we investigate the properties on the mutiensemble sampling (MES) which is one of the solutions that surmount limitations of conventional sampling algorithms. Works for finding out practical systematic ways of using the MES efficiently to explore distantly separated regions in configuration space are performed. In this work, the more generalized form of weighting function for MES is used and 'cavity formation in water' is simulated using Monte Carlo. investigating the correlation of simulation parameters and the efficiency of the method, we propose a practical way of maximizing the power of the MES. We applied the way to 'cavity formation in water' and were able to explore the parts of configuration space relevant to cavities of radius from 0 to 5.6A in a single simulation.