• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.6B
• /
• pp.794-801
• /
• 2001

본 논문에서는 rank-order 필터의 구현을 위해 세 가지 신경회로망의 구조를 제시하고 분석하며 용도를 제안한다. 첫 번째 신경회로망을 이용하여 2-입력 정렬기를 제안하고 이를 이용하여 계층적인 N-입력 정렬기를 구성한다. 두 번째로 입력 신호간의 상대적인 크기 정보를 이용하여 학습 패턴을 구성한 후 역전파 학습 기법을 이용하여 구현되는 순방향 신경회로망을 이용한 rank-order 필터를 구현한다. 세 번째로 신경회로망의 구조의 출력층에 외부 입력으로 순위 정보를 가지도록 하는 rank-order 필터를 순방향 신경회로망을 이용하여 구현한다. 그리고 이러한 제안된 기술들에 대해 확장성, 구조의 복잡도와 시간 지연 등에서의 성능을 비교, 평가한다. 2-입력 정렬기를 이용하는 방식은 확장이 용이하고 비교적 구조가 간단하나 입력 신호들의 정렬을 위해 신경회로망은 순환하는 구조를 가지며 입력 신호의 수에 비례하는 반복 연산 후에 결과를 얻게 된다. 반면에, 순방향 신경회로망을 이용한 rank-order 필터의 구현 방식은 이러한 반복 연산으로 인한 시간 지연을 줄일 수 있으나 상대적으로 복잡한 구조를 가진다.

• Journal of the Korea Society of Computer and Information
• /
• v.19 no.8
• /
• pp.27-34
• /
• 2014

In this paper, we propose a pupil detection method using hybrid projection function and rank order filter. To reduce error to detect eyebrows as pupil, eyebrows are detected using hybrid projection function in face region and eye region is set to not include the eyebrows. In the eye region, potential pupil candidates are detected using rank order filter and then the positions of pupil candidates are corrected. The pupil candidates are grouped into pairs based on geometric constraints. A similarity measure is obtained for two eye of each pair using template matching, we select a pair with the smallest similarity measure as final two pupils. The experiments have been performed for 700 images of the BioID face database. The pupil detection rate is 92.4% and the proposed method improves about 21.5% over the existing method..

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.16 no.7
• /
• pp.1383-1390
• /
• 2012

In this paper, we propose a robust pupil detection method using rank order filter and pixel value difference in facial image. We have detected the potential pupil candidates using rank order filter. Many false pupil candidates found at eyebrow are removed using the fact that the pixel difference is much at the boundary between pupil and sclera. The rest pupil candidates are grouped into pairs. Each pair is verified according to geometric constraints such as the angle and the distance between two candidates. A fitness function is obtained for each pair using the pixel values of two pupil regions, we select a pair with the smallest fitness value as a final pupil. The experiments have been performed for 400 images of the BioID face database. The results show that it achieves more than 90% accuracy, and especially the proposed method improves the detection rate and high accuracy for face with spectacle.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.7
• /
• pp.1564-1570
• /
• 2013

In this paper, we propose a robust pupil detection method using rank order filter and cross-correlation. Potential pupil candidates are detected using rank order filter. Eye region is binarized using variable threshold to find eyebrow, and pupil candidates at the eyebrow are removed. The positions of pupil candidates are corrected, the pupil candidates are grouped into pairs based on geometric constraints. A similarity measure is obtained for two eye of each pair using cross-correlation, we select a pair with the largest similarity measure as a final pupil. The experiments have been performed for 500 images of the BioID face database. The results show that it achieves the high detection rate of 96.8% and improves about 11.6% than existing method.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.46 no.4
• /
• pp.9-15
• /
• 2009

This paper proposes a cancellation algorithm of impulse noise using a rank-order method. The proposed method is a fast and simple algorithm that is composed of two parts. The first part involves noise detection using a fuzzy technique, where an image is divided into RGB color channels. Then every pixel in each color channel is investigated and assigned a probability indicating its chances of being a noise pixel. At this time, the rank order method using a noise-detection mask is utilized for accurate noise detection. Thereafter, the second part involves noise-cancellation, where each noise-pixel value in an image is replaced in proportion to its fuzzy probability. Through the experiments, both the conventional and proposed methods were simulated and compared. As a result, it is shown that proposed method is able to detect noisy pixels more accurately, and produce resulting images with high PSNR values.

• Journal of the Korean Society of Clothing and Textiles
• /
• v.17 no.4
• /
• pp.550-564
• /
• 1993

The purpose of this study is to investigate the consumer's claims related to clothing merchandise. By th origination stage of claims, details of claims, and treatments of claims purchasing places of clothing merchandise, the consumer's claims are analyzed which were lodged to in consumer's complaint center, Seoul YWCA, in 1981-1990. To analyze these data statistically, frequency and percentile are used. The results of analysis for consumer's claims are as next : 1. Concerning the sex distinction, female complainers are more than male complainers. About the age bracket, twenties and thirties are the most numerous. The originations of claims being various. It is laundry and dry cleaning stage out of them that rank first, and total numbers of claims for clothing products continually have increased during 1981-1990. Out of the clothing items, outerwears are of the first rank and formal wear and coat are highest in rank of outerwears. For claims about purchasing places, agency ranked first and market, department store, custome-made and discount store came after in order. 2. Concerning the contents, quality of clothing product ranks first, inferior service, price, contrast, unfair transaction ranks in order. There are claims about quality of clothing product that color change ranks first and damage and form change rank in order. 3. The treatments of claims are that counsel, exchange, refund, repair and correction rank in order.

• The Pure and Applied Mathematics
• /
• v.31 no.2
• /
• pp.159-178
• /
• 2024

We continue the investigations in [7] extending the Bruhat order on n × n alternating sign matrices to a more general setting. We show that the resulting partially ordered set is a lattice and also investigate its rank.

• Journal of the Korean Data and Information Science Society
• /
• v.15 no.3
• /
• pp.617-624
• /
• 2004

In this article the fuzzy number rank and the fuzzy rank transformation method are introduced in order to analyse the non-parametric fuzzy regression model which cannot be described as a specific functional form such as the crisp data and fuzzy data as a independent and dependent variables respectively. The effectiveness of fuzzy rank transformation methods is compared with other methods through the numerical examples.

• Journal of Korean Institute of Industrial Engineers
• /
• v.26 no.1
• /
• pp.27-35
• /
• 2000

For a weighting method to be practically valid, it should produce weights which coincide with the relative importance of attributes perceived by the decision maker. In this paper, 'bootstrapping' is used to compare the practical validities of five weighting methods frequently used; the rank order centroid method, the rank reciprocal method, the rank sum method, the entropic method, and the geometric mean method. Bootstrapping refers to the procedure where the analysts allow the decision maker to make careful judgements on a series of similar cases, then infer statistically what weights he was implicitly using to arrive at the particular ranking. The weights produced by bootstrapping can therefore be regarded as well reflecting the decision maker's perceived relative importances. Bootstrapping and the five weighting methods were applied to a job selection problem. The results showed that both the rank order centroid method and the rank reciprocal method had higher level of practical validity than the other three methods, though a large difference could not be found either in the resulting weights or in the corresponding solutions.

• Journal of applied mathematics & informatics
• /
• v.32 no.1_2
• /
• pp.171-184
• /
• 2014

A higher order iterative method to compute the Moore-Penrose inverses of arbitrary matrices using only the Penrose equation (ii) is developed by extending the iterative method described in [1]. Convergence properties as well as the error estimates of the method are studied. The efficacy of the method is demonstrated by working out four numerical examples, two involving a full rank matrix and an ill-conditioned Hilbert matrix, whereas, the other two involving randomly generated full rank and rank deficient matrices. The performance measures are the number of iterations and CPU time in seconds used by the method. It is observed that the number of iterations always decreases as expected and the CPU time first decreases gradually and then increases with the increase of the order of the method for all examples considered.