농담에서의 형상복구를 이용한 복합물체의 재구성 방법

권영빈 중앙대학교 기술과학연구소 1990 기술과학연구소 논문집 Vol.20 No.-

In this paper, a new method for the reconstruction of composite object is proposed. The conventional shape from shading algorithm does not find out the composite object's boundary information. Using the junction points, the 3D reconstruction of composite object on a single image can be carried out.

한국인 주요우울장애 환자에서 혈중 ACTH, Cortisol 농도와 해밀턴 우울 평가 척도의 신체증상 항목과의 상관관계

여혜빈,김린,함병주,심세훈,권영준,정희연,정한용,한상우,이민수,이화영 대한생물치료정신의학회 2011 생물치료정신의학 Vol.17 No.2

Objectives:Hypothalamic-Pituitary-Adrenocortical(HPA) system dysregulation plays an important role in the pathophysiology of major depressive disorder. Medically unexplained somatic symptoms comprised the predominant complaints of korean patients with major depressive disorder. It might be related to dysregulation of HPA system and somatic symptoms. The aim of this study was to investigate the association between Hamilton Depression Rating Scale severity and the Plasma Adrenocorticotropic Hormone(ACTH) & Serum Cortisol level in Korean patients with major depressive disorder. Methods:Our study design was prospective. A total 111 hospitalized patients with major depressive disorder (DSM-IV) at Anam Hospital, Korea University Medical Center between Jan 2009 and May 2011 were selected. Hamilton Depression Rating Scale was measured by psychiatrist. Plasma ACTH, Serum Cortisol levels were analyzed at the time of admission. Results:There are significant correlations(p<0.05) between Hamilton Depression Rating Scale of somatic component(HAM-D 11-15 item) and Plasma ACTH and Serum Cortisol level. However, other HAM-D sub items were not significantly correlated with Plasma ACTH and Serum Cortisol level except somatic component. Conclusion:Our Study results suggest that hyperactivity of HPA axis might be correlation with somatic symptoms in korean patients with major depressive disorder.

온라인 한글 인식에 적합한 새로운 신속 정합 알고리즘

권영빈,권오성 중앙대학교 생산공학연구소 1993 생산공학연구소 논문집 Vol.2 No.1

일반적으로 한글을 포함한 문자 인식은 문자를 이루는 획수와 형태의 심한 변형에 따른 어려움을 갖고 있다. 본 논문에서는 이러한 변형을 효과적으로 처리할 수 있는 스트링 정합 방법을 제안한다. 이 방법에서 입력되어진 문자는 선분들의 방향값과 특징에 따라 정의된 기호들의 스트링으로 변환된다. 스트링 사이의 정합은 스트링 편집과 비교순서함수에 의해 이루어 진다. 스트링 편집은 교환, 삽입, 삭제 연산에 의해 행해지고, 비교순서함수는 연산 결과에 따라 다음에 비교할 스트링의 기호 순서를 정의한다. 인식 실험을 통하여 제안하는 방법은 스트링 정합방법으로 자주 사용되는 DP(dynamic programming)방법에 비하여 1/18 정도의 수행시간을 요구하고, 50명이 쓴 21,076자의 필기에 대하여 88.96%라는 만족스런 인식률을 제공하였다. In general, the recognition of characters including Hangul has been many troubles owing to the severe variations of stroke number and of their shape. In this paper, we propose a method efficiently handling the variation of handwriting styles. An input characetr is converted into the line-segments of equal size and they forms the string after extracting the features. A Matching between strings is carry out by a string editing and the order of the comparison is done by the predefined function. The string editing is defined by the operations such as change, insert, and delete. Using the result of string editing, the predefined function for comparison order defines the orders of symbols in a string to be compared. From experimental results, the proposed method requires only 1/18 of recognition time comparing with the DP method and it reaches 88.96% of recognition rate over 21,076 cursive characters obtained from 50 different persons.

감염질환에 의한 변사자의 장기조직에서 결핵균의 검출에 관한 연구

이한영,엄용빈,최철호,김영주,권태정 內務部 國立科學搜査硏究所 2001 國立 科學 搜査 硏究所 年報 Vol.33 No.-

윤곽선 스펙트럼에 의한 인쇄체 문자인식

김민기,권영빈 중앙대학교 생산공학연구소 1996 생산공학연구소 논문집 Vol.5 No.1

윤곽선은 다양한 정보를 포함하고 있기 때문에 윤곽선으로부터 중요한 인식 특징을 찾고자 하는 많은 연구방법들이 개발되었다. 본 논문에서는 윤곽선의 샘플링과 양자화에 의한 새로운 문자인식 방법을 제안한다. 본 논문의 기본 아이디어는 윤곽선의 다차원 정보를 일차원 정보로 분해하는 것이다. 윤곽선을 추적하면서 수평, 수직, 그리고 각도의 변이를 추출하여 일차원 정보를 갖는 세 개의 윤곽선 스펙트럼을 구성하게 된다. 그러나 이러한 윤곽선 스펙트럼은 윤곽선 추적의 시작점, 글꼴의 유형, 크기에 민감하기 때문에 윤곽선 추적의 시작점을 동일하게 유지시킬 수 있는 방법을 제시하였다. 그리고 샘플링과 양자화를 통해 글꼴의 유형과 크기에 무관한 윤곽선 스펙트럼을 생성할 수 있다. The contour of character includes many different features. Numerous approaches are explored to find essential features which are extracted directly from the contour. In this paper, we propose a new character recognition method based on the sampling and quantization of contour information. Our idea is to decompose the multidimensional features of a contour into one-dimensional contour spectrums. Following the contour, we extract horizontal, vertical, and angular variations and make three one-dimensional contour spectrums from these features. These contour spectrums are sensitive to the starting point of contour tracing, and it varies on the font type and size, so we propose a novel algorithm which finds the invariant staring point of contour tracing. By sampling and quantization, we can make discrete contour spectrums which are invariant to the font and size.

트리 분류기를 이용한 한글 문자의 유형 분류

김민기,권영빈 중앙대학교 생산공학연구소 1997 생산공학연구소 논문집 Vol.6 No.1

일반적으로 한글은 6가지 유형으로 분류하고 있는데 이러한 유형 분류는 한글을 음절 단위로 인식할 때는 대분류의 효과를 갖게 되며 자소 단위로 인식할 때는 자소 분할을 효과적으로 수행하기 위한 전단계의 역할을 하게 된다. 본 논문에서는 글꼴의 변화와 잡영을 흡수할 수 있는 초성, 중성, 종성의 자소 영역을 정하여 이 영역에 나타나는 횡모음과 종모음의 주획을 추출하는 방법을 기술하였다. 종모음 영역에서 추출한 수직획들과 횡모음 영역에서 추출한 수평획들을 각각 종모음과 횡모음의 후보들로써 이들로부터 종모음과 횡모음을 올바로 추출하는 것이 한글 유형 분류의 핵심적인 내용이다. 그러나 수평획들로부터 횡모음을 바로 추출하는 것은 매우 어려운 문제이기 때문에 트리 분류기를 이용하여 복잡하고 다양한 특징을 단계별로 단순화시키고 트리 분류기의 상위 노드에서 결정된 정보와 제약 조건을 이용하여 횡모음을 추출하는 방법을 제안하였다. In general, Hangul is clustered into six types. The type clustering has an effect on coarse classification in syllable matching and prior step for grapheme segmentation in grapheme matching. In this paper we define a set of grapheme regions which include a part of graphme; pre-consonants, vowels, post-consonant. Each region is robust in various fonts and noises. We described a method which extracts a main stroke of vowel touched on each region. Extracting main stroke of vowel among horizontal strokes or vertical strokes is an essential topic in 6-type Hangul clustering. However, choosing a h-vowel(vowel of horizontal direction) among horizontal strokes is very difficult. We simplified complex and various features using a tree classifier and proposed an h-vowel extraction method based on the information and constraints acquired from upper nodes of tree classifier.

원고지 상에 쓰여진 온라인 필기 제스쳐 인식기 구현

정우식,권영빈 중앙대학교 정보통신연구소 2000 정보통신연구소논문집 Vol.2 No.2

본 논문에서는 원고지 상에 작성된 문자들의 교정을 위한 제스쳐 인식 방법을 제안하고 있다. 원고지의 교정용으로 제시된 13개의 정의된 제스쳐를 중심으로 흘려 쓴 원고의 내용을 수정할 수 있도록 하였다. 이때, 인식을 위하여 제시된 프리미티브로는 직선, 곡선 및 원을 사용하였으며 이들은 벡터 세크먼트들의 조합에 의한 결합과정에 따라 한글 오토마타에 의해 정의되도록 하였다. 제시된 방법의 인식율은 40명의 다른 필기자가 작성한 520개의 샘플에 대하여 92%를 나타내어 실용화가 가능함을 보여주고 있다. In this paper, a method on handwriting gesture recognition for the oriental manuscript paper form is proposed. The pre-defined 13 different kinds of gestures are introduced on the boxed type manuscript paper on which the cursive handwrithing characters are already written. To detect the pre-defined gestures, primitives of line,curve, and circle are defined from the pre-calculated vector segments and its calculation using line merging process and automata. The recognition ratio of the proposed method is obtained at 92% among 520 samples with 40 different writers.

어린이 김치 표준화에 대한 연구(1) : 어린이 김치의 담금법 개발 Development of the Preparation of Kimchi for the Children

송영옥,전영수,권명자,빈성미,김은희,문정원,김명 부산대학교 가정대학 1995 家政大學硏究報告 Vol.21 No.-

The purpose of this study is reporting the results of first trial for the standardization of children's kimchi in terms of brining standardization, development for the recipe ad pannel test for the children's kimchi. In order to develop the standard recipe for children's kimchi, our research team have visited kimchi factories, temples and houses located in Pusan and Kyung-Sang Nam Do province. Three kinds of chinese cabbage kimchi (hamyang, haeundae, and jinrae kimchi named by region where we collected) which got the top three highest score in panel testing was selected. The original recipes of these kimchi were sightly modified to prepared the kimchi at laboratory. Among three of them, hamyang kimchi which contained dried anchovy power was selected as a model kimchi for development of the children's kimchi recipe. We find out that the optimal salt concentration of brined cabbage and kimchi for the children's kimchi (sliced style kimchi, approximately 2.5X3cm a piece) that showed the hightest pennel score were approximately 1.5% and 2.0%, respectively. The salt concentration of brined cabbage became 1.5, 1.43, and 1.60% after 60 min at 10% brining solution, 45 min at 12% and 30 min at 15%, respectively. The developed recipe used for the children's kimchi at our team is as follows. The major ingredients used for the preparation of kimchi is brined sliced chinese cabbage (salt concentration is 1.5%) 1kg galic 11g, ginger 5.0g, red pepper powder 20.0g, fermented anchovy juice 25.0g, fermented shrimp juice 5.0g dried anchovy and sea weed extracts 30.0g, anchovy power 2.4g glutinious rice gel 40.0g, starch syrup, 10.0g, and apple juice 22.5g. The total acceptability test of developed children's kimchi (test kimchi) was carried out with 84 of 6th grade student during lunch time at elementary school in Pusan where has a school lunch program. Sixty nine percent of student answered test kimchi is very good, 25% of them said it is fair, and only 6% of students answered they didn't like test kimchil. However, for the kimchi supplied from school at lunch, 52% students answered the kimchi is good, 26% of them said it is fair, and 22% of them said they don't like school supplied kimchi. This result tells us that the test kimchi is better than school supplied kimchi and the test kimchi can be developed further as a children's kimchi.

운동기간에 따른 혈중지단백대사 및 비만호르몬의 변화

홍성찬,함용기,김영빈,박병근,정동혁,이정윤,권성택 圓光大學校 附設 體力科學硏究所 2003 體力科學硏究 Vol.26 No.2

The purpose of this study is to elucidate the effects of regular swimming exercise on blood lipoprotein metabolism and leptin. In this study, swimming(HRmax 50-75%, 60min/day, 5day/week) was performed and 18 were participated for study during 16 weeks. They were divided into two groups, one group as experimental group(n=9) and the other as control group(n=9). A measured factors were TC, TG, HDL-C, LDL-C and leptin.. All data were expressed as mean± standard deviation by using spss package program(win 9.0), one way anova and independent t-test was used to analyze the difference of exercise duration(0week, 8week, 12week, 16week) between groups. For all statistics performed statistical significance was set at p<.05. After exercise intervention, TC, TG, HDL-C, LDL-C and leptin were found to have more significant effects in the experimental group than control group. This results indicated regular swimming exercise training is beneficial to prevent and treat cardiovascular disease and obesity.

Poster Session : Expression and biochemical properties according to fusion tag position of human tyrosinase without transmembrane domain in bacterial system

( Young Bin Kwon ),( Kwang Hoon Kong ) 한국생화학분자생물학회 (구 한국생화학회) 2007 생화학분자생물학회 춘계학술발표논문집 Vol.2007 No.-