表形式文書에서 틀情報와 접촉된 文字의 固定度 抽出에 관한 硏究

金東炫 順天靑巖大學 1996 論文集-順天靑巖大學 Vol.20 No.1

In this paper, we propose new method of high accurate extracting form information and touching characters from digitized images of filled-out forms. Long and short Run-Length Groups are formed into connected components which are filtered through threshold values. Then, line elements are extracted from horizontal and vertical lines in the groups. The broken characters after removing touching lines segments are recovered using patching methods - arc patching, quadrilateral patching, double quadrilateral patching, binding arc patching, and double arc patching. This technique allows the high accuracy extraction of characters. About filled out forms of the gross of 11 numbers, results of experiments by methods of realization in this paper, it could gain that 97.2% of recovering rates is combination of 417 characters except 12 characters in the 429 charact- ters witch are the character gross number, and then, shows the effectiveness.

現行原價會計의 制度的 定着

金東律,趙顯玉 釜山外國語大學校貿易經營硏究所 1987 貿易經營論集 Vol.2 No.-

Persistent inflation in 1970s made the relevancy of disclosed traditional accounting data thoroughly suspected. As a result, some actions for developing an alternative accounting system to reflect the impact of inflation on the financial statements reported by business entities are required. Two basic approaches for inflation-adjusted accounting system are suggested; one is the current purchasing power accounting(CPP) which use the general price index numbers to reflect the change of purchashing power of money according to the impact of inflation, the other is current cost accounting(CCA) which use current cost instead of historical cost to assess the value of resources. In the early 1970s most of official recommendations were concerned with general purchasing power accounting, and in 1975 it seemed probable that the accountancy bodies in several countries would be recommending the publication of supplementary CPP accounting information in the late 1970s. In September 1975, however, the UK Sandilands Committee recommended a system of current value accounting, and since that time there has been a gradual movement away from CPP accounting towards current cost accounting. In the paper, the reasons of the reluctance to accept the inflation-adjusted accounting system are revealed. And the theoretical background for the broad acceptance of CCA instead of relatively easy and cost-saving CPP accounting are also examined. Moreover, the need for and the possibility of the introduction of CCA to Korean economy are diacussed, and the proposals for CCA acceptance are offered as a final comment.

大學의 새로운 役割과 産學協同

金東炫 漢陽大學校 行政問題硏究所 1996 行政問題論集 Vol.14 No.-

전자상거래에서의 조정 에이전트 시스템에 관한 연구

金東炫 順天靑巖大學 2001 論文集-順天靑巖大學 Vol.25 No.-

At present, most Electronic Commerce System is made mainstream of Server-Client Commerce. For searching goods information, we must entering the EC sites and search that's information But, P2P System being escape traditional structure, Server-Client, it enable information transact between users, and resolve the problem of network, add-cost of server-ensurance, etc. Also, it is possible to trade information between person to person, client to provide. In this paper, we analysis the model of EC business progress presently. One of the features of the proposed mechanism is with improved agent route table is supply better transparency in finding out agent locations, better recoverable of agents, and the ability of detecting activities of deliberate, illegal copy.

1992학년도 간편 MMPI 성격검사 결과분석

金東規,朴賢珍 釜山工業大學校 學生生活硏究所 1993 學生生活硏究 Vol.3 No.-

Neural Network를 이용한 혼용 문서 인식에 관한 연구

金東炫 順天靑巖大學 1997 論文集-順天靑巖大學 Vol.21 No.1

In rhia paper, we propose the document recognition system that can recognize the multi-size, multi font and mixed type characters. We have utilized a large scale neural network model, "CombNET" which consists of a 4 layered network with a comb structure. And we proposed recognition method that can recognize characters without discrimination of character type. The first layer constituts a Kohonenen's SOFM network which quantizes an input feature vector space into several sub-spaces and the following 2-4 layaers constitutes BP network modules which classify input data in each sub-space into specified categories. An experimental result demonstrated the usefulness of this approach with the recognition rates of about 95.1 for the training data and 93.8 for the test data. For the mixed type character documents we obtained the recognition rates of about 95. 1 and recognition speed of 10.4 characters per second, and then, shows the effective ness.

멀티캐스트에서의 데이터 보완 시스템에 대한 연구

金東炫,許賢 順天靑巖大學 2001 論文集-順天靑巖大學 Vol.25 No.-

The internet has the advantage of connecting to a server conveniently any time, any where by means of open networks. By the way, its problems of data security have emerged as a major concern for the purpose of activity in growth performing electronic transaction and the intranet etc., because there coexists the possibility of intrusion of hackers and forgery or falsification of data owing to the unreliability of protecting confidentiality. There are solutions to this problem. One of them is the method of using PGP that is applied to the e-mail security. This method is one of the most widely used. However, there is a major defect that it takes too much time to transmit a message because there are increased traffic among the application programs, encryption modules, and the browsers. This defect comes mainly because the encryption module stays outside of the WWW system. We used the information security system devised for the secure WWW data transmission by extending the functions of the Netscape browser and by using application programs such as Java applications and by using the plug-in methods.

오류특징점 제거 알고리즘을 이용한 화상 인식시스템에 관한 연구

金東炫,정양권 順天靑巖大學 1998 論文集-順天靑巖大學 Vol.22 No.1

In this paper, we proposed an algorithm to eliminate false feature points to upgrade the system ability using the cross number method. By using the proposal algorithm, it was able to extract feature points exactly and increase the recognition rates. In the experiment, we use 3 groups : the cut, the uncut-partially noised, and uncut completely noised fingerprint to verify our proposal. We then compared the results of the proposed system with those of another system in which false feature points are not eliminated. We have obtained good recognition rates of 97.9%, 97.9%, and 95.1% in the cut, uncut-partially noised, and the uncut-completely noised fingerprints. However, the other system received the results of the cut 60.%, the uncut- partially noised 35.3%, and the uncut-completely noised 50.2% respectively. As a result, we believe that the fingerprint may be recognized after adapting the proposal algorithm.

승온시 Si₂H?가스 주입을 이용한 고상에피택시에 의해 제조된 Si에피택시막에 관한 연구

金東玄,申東勳,金世埈,金亨駿 弘益大學校 科學技術硏究所 1997 科學技術硏究論文集 Vol.8 No.-

In this study, a novel technique to realize a selectively grown Si epitaxy layer onto source/drain regions has been developed. This technique consists of deposition of amorphous Si with oxide-free Si surface, selective solid phase epitaxial(SPE) growth on Si windows. The amorphous Si layer serves as the channeling prevention layer during the ion implantation. Usually, amorphously deposited Si films by CVD are crystallized to poly-Si due to a formation of SiO₂layer at amorphous Si/Si substrate interface. Formation of SiO₂layer on si surface can be effectively prevented by flowing Si₂H?? gas during the heating-up procedure fro amorphous Si depositions. In this way, amorphously deposited Si layer onto (100) Si substrates was grown epitaxially during the post-deposition heat treatments. Employing Si₂H?? flowing and solid phase epitaxial growth, high-quality epitaxial Si layer was obtained at low temperatures below 600℃ without conventional high temperature cleaning procedures.

官僚制의 成立과 發展過程에 관한 硏究

金東炫 中央大學校 法政大學 學生會 1965 法政論叢 Vol.11 No.1