에지 분석에 의한 자동 독화 실험

이경호,금종주,이상범,Lee, Kyong-Ho,Kum, Jong-Ju,Rhee, Sang-Bum 한국컴퓨터산업학회 2008 컴퓨터産業敎育學會論文誌 Vol.9 No.1

In this paper, the edge parameters were drawn from speaking image around lip and effective automatic lip reading system to recognize the Korean 'a/e/i/o/u' 5 owels were constructed using the parameter. Speaking images around lip were divided into $5{\times}5$ pane. In each pane the number of digital edge element using Sobel operator were evaluated. The observational error between samples was corrected by using normalization method and the normalized value is used for parameter In the experiment to convince the strength of parameter, 50 normal persons were sampled. The images of 10 persons were analyzed and the images of another 40 persons were experimented for recognition. 500 data are gathered and analyzed. Based on this analysis, the neural net system is constructed and the recognition experiments are performed for 400 data. The neural net system gave the best recognition result of 91.1%.

<<한국어 5모음의 조음적 제어 분석을 이용한 자동 독화에 관한 연구>>

이경호,금종주,이상범,Lee, Kyong-Ho,Kum, Jong-Ju,Rhee, Sang-Bum 한국컴퓨터산업학회 2007 컴퓨터産業敎育學會論文誌 Vol.8 No.4

In this paper, we set 6 interesting points around lips. Analyzed and characterized is the distance change of these 6 interesting points when people pronounces 5 vowels of Korean language. 450 data are gathered and analyzed. Based on this analysis, the system is constructed and the recognition experiments are performed. In this system, we used the camera connected to computer to measure the distance vector between 6 interesting points. In the experiment, 80 normal persons were sampled. The observational error between samples was corrected using normalization method. We analyzed with 30 persons and experimented with 50 persons. We constructed three recognition systems and of those the neural net system gave the best recognition result of 87.44 %.

지리도면의 자동 벡터화를 위한 영상 세선화 알고리즘의 체계적인 성능평가

이경호,김경호,조성배,최윤철,Lee, Kyung-Ho,Kim, Kyong-Ho,Cho, Sung-Bae,Choy, Yoon-Chul 한국정보처리학회 1997 정보처리논문지 Vol.4 No.12

최근에 사회 여러 분야에 걸쳐서 지리정보를 효과적으로 저장하여 검색할 수 있도록 하는 지리정보시스템에 대한 관심이 고조되고 있다. 이를 구축하는데 가장 많은 시간과 비용이 소요되는 부분은 지리정보의 입력 부분으로 효율적인 입력 방법을 선택하는 것이 매우 중요하다. 여기에서 입력 영상으로부터 한 화소 두께의 골격선을 추출하는 영상 세선화 알고리즘은 전처리 단계로서 필수적이라 할 수 있다. 본 논문에서는 그동안 주로 문 자 인식 등의 영상처리를 위해 발표된 영상 세선화 알고리즘들 중에서 대표적인 알고리즘을 등고선도와 지적도 및 상하수도 등의 지리도면에 적용하여 비교 분석함으로써 각각의 지리도면의 자동 벡터화에 적합한 알고리즘을 제시하고자 한다. 여러가지 평가 기준에 의해 실험한 결과, 등고선도의 경우에는 Arcelli의 방법, 지적도는 Holt의 방법, 그리고 상하수도는 Chen의 방법이 각각 적합하다는 것을 알 수 있었다. In a variety of fields, recently, there is a growing interest in Geographic Information System which facilitates efficient storage and retrieval of geographic information. It is of extreme importance to make a good choice of efficient input method, because it takes the most of the lime and cost in constructing a GIS. Among several steps, thinning input image to produce skeleton of unit width is prerequisite to the automatic input or geographic maps. In this paper, we systematically evaluate the performance of representative thinning algorithms in geographic maps such as contour, cadastral, and water and sewer maps, and suggest appropriate algorithms for the maps, respectively. A thorough experiment indicates that Arcelli's method is best for contour maps, Holt's method for cadastral maps, and Chen's method for water and sewer maps.

대입문 병합을 이용한 효율적인 자바 난독화 기법

이경호(Kyong-Ho Lee),박희완(Hee-Wan Park) 한국컴퓨터정보학회 2013 한국컴퓨터정보학회논문지 Vol.18 No.10

Java bytecodes are executed not on target machine but on the Java virtual machines. Since this bytecodes use a higher level representation than binary code, it is possible to decompile most bytecodes back to Java source. Obfuscation is the technique of obscuring code and it makes program difficult to understand. However, most of the obfuscation techniques make the code size and the performance of obfuscated program bigger and slower than original program. In this paper, we proposed an effective Java obfuscation techniques using assignment statements merging that make the source program difficult to understand. The basic approach is to merge assignments statements to append side effects of statement. An additional benefit is that the size of the bytecode is reduced.

DSSSL에 기반한 SGML 문서편집기의 설계 및 구현

이경호(Kyong Ho Lee),고승규(Seung Kyu Ko),백승욱(Seung Uk Baek),변창원(Chang Won Byun),장은영(Eun Yeung Chang),최윤철(Yoon Chul Choy) 한국정보과학회 1998 정보과학회 컴퓨팅의 실제 논문지 Vol.4 No.6

SGML은 이 기종간에 호환이 가능하고 논리적인 구조 정보를 포함한다는 장점 때문에 CALS, EC, 인트라넷/그룹웨어, 전자도서관 등의 다양한 분야에서 문서처리의 표준 포맷으로 자리잡았다. 이에 SGML 문서를 쉽게 생성 및 편집할 수 있는 문서편집기에 대한 요구가 급증하고 있으며 현재 전세계적으로 이에 대한 연구가 활발히 진행 중이다. 그러나 기존의 문서편집기는 SGML 문서의 프리젠테이션을 위하여 독자적인 스타일 포맷을 사용한다. 이는 개방형 문서 표준이라는 SGML의 본래 취지에 어긋나는 것이다. 이에 ISO/IEC에서는 SGML 문서의 변환 및 스타일 정보를 기술할 수 있는 표준 안으로 DSSSL을 제안하였다. 따라서 본 논문에서는 DSSSL을 지원하며 한글처리가 가능한 SGML 문서편집기를 개발하였다. 이를 위하여 한글을 지원하는 DSSSL 처리기의 표준적인 인터페이스를 지원하기 위하여 Grove 생성기를 개발하였으며 SGML 문서의 프리젠테이션 결과를 출력할 수 있는 포맷터를 개발하였다. Due to the fact that SGML is independent of the platform and is a means for embedding logical structure information into documents, it has been widely accepted as a standard format for representing documents in various fields such as CALS, EC, Intranet/Groupware, and Electronic Library. There are growing interest and studies in an SGML document editor that facilitates creation and editing of SGML documents. Existing editors use vendor-dependant style format for formatting documents. However, it goes against the original purpose of SGML that is called an open standard. For this reason, recently, ISO/IEC proposed DSSSL as an international standard for associating processing information, transformation and formatting, with SGML documents. Therefore, we design and implement an SGML document editor, which can process DSSSL style sheet and Hangul. Specially, m order to implement the proposed editor, we develop an SGML parser that can process Hangul, and a DSSSL processor that deals with DSSSL style language. And we develop a Grove building processor which supports a standard interlace between an SGML parser and a DSSSL processor, and a formatter which can display the result of formatting an SGML document.

블록 피하기 게임에서 강화 학습을 이용한 블록 피하기 에이전트 구현

이경호(Kyong-Ho Lee),강병섭(Byong-Seop Kang) 한국컴퓨터정보학회 2018 한국컴퓨터정보학회 학술발표논문집 Vol.26 No.1

본 논문에서는 2차원 공간상에서 상부에서 하부로 떨어지는 블록을 하부에서 피하는 게임에서 강화 학습에 사용되는 DQN 알고리즘을 이용하여 블록 피하기 에이전트를 구현하고 학습 통해 점점 더 높은 점수를 받는 모습을 확인하였다. 파이썬을 이용하여 게임을 구현한 후 텐서플로우를 이용하여 DQN를 이용한 에이전트를 구현하였다. 에이전트는 보상을 통한 학습을 통하여 점점 강화되도록 하였는데, 초기에는 무작위로 움직였으나, 환경으로부터 받는 보상으로 점점 더 능숙하게 피하는 모습을 관찰할 수 있었다. 본 구현에서는 4000번 정도의 게임 시행에서 아주 능숙하게 피하는 결과를 얻을 수 있었다.

색상과 얼굴 특징 정보를 이용한 얼굴 추적

이경호(Kyong-Ho Lee) 한국컴퓨터정보학회 2013 한국컴퓨터정보학회논문지 Vol.18 No.11

TIn this paper, we find the face in color images and the ability to track the face was implemented. Face tracking is the work to find face regions in the image using the functions of the computer system and this function is a necessary for the robot. But such as extracting skin color in the image face tracking can not be performed. Because face in image varies according to the condition such as light conditions, facial expressions condition. In this paper, we use the skin color pixel extraction function added lighting compensation function and the entire processing system was implemented, include performing finding the features of eyes, nose, mouth are confirmed as face. Lighting compensation function is a adjusted sine function and although the result is not suitable for human vision, the function showed about 4% improvement. Face features are detected by amplifying, reducing the value and make a comparison between the represented image. The eye and nose position, lips are detected. Face tracking efficiency was good.

전자책 포맷 기술 및 표준화

이경호(Kyong-Ho Lee),임순범(Soon-Bum Lim) 한국정보과학회 2010 정보과학회지 Vol.28 No.10

분할 가중치 테이블 역전파 신경망을 이용한 구구단 학습 기능성 게임 제작에 관한 연구

이경호(Kyong-Ho Lee) 한국컴퓨터정보학회 2009 한국컴퓨터정보학회논문지 Vol.14 No.10

In this paper we made the serious game learning multiplication table to be evolved. The serious game is to induce the interest of the learner. This program has an interaction form which reflects the intention of the user and using this program a learner to learn the multiplication table as teacher's location are training a program that are seen as the abata and came to be that learner is smart. A study ability to be evolved used an back propagation neural networks. But we improved a study speed using divided weight table concept. The engine is studied perfectly in 60~80 times training. The learning rate increase to various non-monotonic functional form not to do a mechanical rise. And the learning rate is similar with the study ability of the human.

사각형 개수로에서의 수차 흘수 변화에 따른 회전 효율 파악을 위한 시뮬레이션

이경호(Kyong-Ho Lee),박희완(Hee-Wan Park) 한국컴퓨터정보학회 2013 한국컴퓨터정보학회논문지 Vol.18 No.4

In this paper, simulations were carried out to determine the efficiency of the rotation efficiency according to the draft of waterwheel in open rectangular channel. In the small hydroelectric generators to get the highest efficiency of waterwheel is very important. But the presence of various elements(free water surface flow, non-uniform velocity distribution because of the waterways wall friction etc) makes it difficult to create a mathematical formula. In this paper, we made a scale model and perform a physical simulation where the draft, gradient and flux is variable. Scale modelling with 10-step draft, 3-step gradients and 2-step flux, as well were constructed then computerized automatic experimental system were configured to acquire the rotational efficiency vs. draft of itself. Rotational efficiency is analyzed as for the draft of waterwheel using the acquired data by varying the gradient and flux of canal. Reviewing the analyzed data, it is confirmed that phenomena of efficiency shown at previous and present experiment is similar and revealed that computerized system shows more sophisticated numerical figures.