철판 2매 검출용 비젼시스템의 개발

심재창,권찬영 安東大學校 工業技術硏究所 1998 공업기술연구논문집 Vol.5 No.1

본 논문에서는 컴퓨터비젼과 레이저 조명을 이용하여 프레스에 투입되는 철판의 매수를 감시하는 시스템을 개발하고 인식 알고리즘을 제안하였다. 제안된 시스템은 조도, 진동 및 주변 환경이 열악한 현장에 적용할 수 있도록 조명으로 빔 레이저를 사용하였다. 약 1/4초 이내에 처리가 가능한 고속 인식 알고리즘을 구현하였다. 철판 매수 확인을 위해서는 철판 한 장의 평균 두께의 분포를 구하고 이 정보를 이용한다. 제안된 시스템을 자동차 부품을 제작하는 프레스의 앞단에 설치하여 운영해 본 결과, 철판이 프레스로 입력되기 직전에 두 장 이상의 철판이 겹쳐져서 입력되는 것을 사전에 방지할 수 있었다. In this paper, we developed an automatic steel plates counting system for a huge press, automobile parts production machine, using computer vision technique with beam laser. And we also proposed a fast recognition software algorithms. The steel sheet must be fed into the press only one sheet at a time. In case, if it is fed into two or more then the press will be broken. The proposed system has a new lighting system employing line beam laser. It can capture nearly same images in any condition such as day or night, with light or not, and vibration or not. Our recognition algorithm takes only 0.25 second per cycle. The average of steel thickness is used for counting the sheets of steel. We have installed the proposed system in front of the press machine in a factory. As a result, our system can prevent feeding into two or more steel sheets accidently.

Pascal 프로그램을 위한 Syntax Tree의 설계및 구현

심재찬,이만호 충남대학교 자연과학연구소 1993 忠南科學硏究誌 Vol.20 No.1

Program transformation in source level is a base for parallel programming environment which helps programmer make good parallel programs. Abstract syntax tree is very good intermediate representation of source program to keep the syntax structure of source program. We design and implement the abstract syntax tree for Pascal program, and show the designed abstract syntax tree is good to use as an intermediate representation of Pascal source program by generating Pascal program with same syntax as source program from abstract syntax tree.

진행성 위암과 동반된 후복막 섬유화증 7예의 임상적 고찰

장재영,천영국,최우봉,김진오,조주영,이준성,이문성,심찬섭 순천향의학연구소;Soonchunhyang Medical Research Institute 2000 Journal of Soonchunhyang Medical Science Vol.6 No.1

Background/Aims: Retroperitoneal fibrosis is a fibrosing disease process which frequently develops ureteral obstruction. Malignant retroperitoneal fibrosis has been reported that various types of cancer including cancer of the breast, stomach, prostate, lung, cervix, uteri, colon, pancreas, ovary, and even Hodgkin's disease. While the pathogenesis remains obscure, small foci of metastatic neoplasm in the retroperitoneal space can elicit a desmoplastic reaction in secondary form to malignancy. Methods: Recently we experienced 7 cases of retroperitoneal fibrosis with advanced gastric cancer. So, we report 7 cases of retroperitoneal fibrosis with clinical and radiologic characteristics. Results: The most common signs and symptoms were related irreversible renal dysfunction. The urologic finding noted medial deviation of the involved ureters and hydronephrosis Conclusions: Clinical manifestations are abdominal pain, back pain, hydronephrosis, and uremia. When sudden back pain and hydronephrosis develop in advanced malignancy, physicians should be considerd for retroperitoneal fibrosis.

겔여과크로마토그라피에 의한 질트리코모나스의 항원분획 분석

임미혜,심재찬,민득영,안명희,류재숙,최영길 한양대학교 의과대학 1993 한양의대 학술지 Vol.13 No.1

This study was aimed to analyze the antigenicity of T. vaginalis using a high performance gel filtration chormatography(HPLC), enzyme-linked immunosorbent assay (ELISA) and enzyme-linked immunoelectotransfer blot (EITB). Soluble antigen was prepared using a bead beater. Antiserum was collected from rabbit immunized previously. The results obtained are as follows: the crude antigen of T.vaginalis was separated into 7 peak by HPLC and the fraction 3 revealed the highest reactivity in DLISA. The immunogenic bands of fraction 3 were visualized at 38, 45, 47, 55, 64, 78, 94, 115, 126 and 180 KDa and the bands at 38KDa and 180KDa showed stronger reactivity than that of equivalents of the crude antigen in EITB. In conclusion the fraction 3 of the crude antigen of T. vaginalis obtained by chromatographic separation may have a high antigenicity and proteins ay 38KDa and 180KDa of the fraction 3 may play an important role in antigen-antibody reaction in T. vaginalis infection.

용융염 전해정련 계에서 초 우라늄 대용물질로서 cerium dysprosium의 적용

강영호,황성찬,심준보,유재형 한국공업화학회 2003 공업화학 Vol.14 No.3

회토류원소 중 cerium과 dysprosium은 악틴족 원소와 염화물 생성 자유에너지의 차이가 크지 않기 때문에 분리하기가 매우 어렵다. 즉, 용융염을 사용 하는 전해정련 공정에서는 이들 회토류 원소가 악틴족 원소와 함께 석출되는 특성을 나타낸다. 이러한 성질을 이용하여 본 연구는 용융염을 전해질로 이용하는 전해정련실험에 cerium과 dysprosium금속을 초 우라늄원소의 대용 물질로 사용하여 그들의 모사 특성을 고찰하었다. 실험 결과 용융염 중 낮은 염화우라늄농도에서는 이 두 회토류 원소가 초 우라늄과 유사한 전해 특성을 나타냈으나, 높은 염화우라늄 농도에서는 회토류 원소의 거동을 보여주었다. The rare earth elements, especiaily cerium and dysprosium are very difficult to be separated from actinide elements due to their narrow differences in the Gibbs free energy of formation of chloride. That is, they are codeposited with actinide elements in electrorefining process with molten salt. Using this characteristics, present study discusses the feasibility for use of cerium and dysprosium metals to simulate the electrorefining characteristics of the transuranium. The experimental results show that the electrolytic behaviors of these two rare earth elements are similar with the transuranium in low concentration range of uranium chloride in molten salt. However, in high concentration range of uranium chloride they behave as the rare earth elements.

과학기술계산을 위한 전자계산기용 특수 Library 개발에 관한 연구

金相萬,姜承業,沈在洪,文東纘,徐光烈,洪勝弘,愼哲宰 광운대학교 1972 論文集 Vol.2 No.-

In this paper the subroutine library of digital computer techniques to the solution of electronic networks problem are discussed. The application of digital-computer techniques to the solution of the integrations relating the terminal variables for linear time invariant inductos and capaciters are analyzed, then on the techniques of numerical integrations trapezoidal method is used. The subroutine library for differential equations and determinations which are used in electronic networks analysis have been developed. Since these programs are the general form, anybody can utilize the programs for his own purpose.

Outcomes after the Liver Resection of Colorectal Cancer Liver Metastases: A Single Center Experience

( Jae Ryong Shim ),( Sang Jae Lee ),( Seung Duk Lee ),( Min Jung Kim ),( Sung Chan Park ),( Seong Hoon Kim ),( Sung-sik Han ),( Sang Jae Park ),( Jae Hwan Oh ) 대한간학회 2018 춘·추계 학술대회 (KASL) Vol.2018 No.1

Aims: Colorectal cancer (CRC) is associated with frequent distant metastases. Traditional surgical option for colorectal liver metastasis (CRLM) was a staged operation. And with advances of surgical techniques, devices, perioperative management and chemotherapeutic agents, CRLMs are now believed to be curable by operation. We analyze the differences of perioperative and oncologic outcome between simultaneous and staged operation. Methods: Four hundred fifty six patients who underwent hepatic resection for known CRLM between January 2001 and December 2014 were retrieved from a retrospective database at our institution. Simultaneous resection was defined as co-operation between colorectal resection and liver resection. And staged operation was defined as colon resection was performed first and additional chemotherapy was followed. Results: There were no statistically significant differences except ASA, preoperative chemotherapy state, number of metastatic tumor, liver resection margin (P=0.034, P=0.001, P=0.001, and P=0.017, respectively). Only major liver resection showed statistically significant difference in multivariate analysis for postoperative complication (P=0.018). In the multivariate analysis for disease-free survival and overall survival, there was a statistically significant increased risk of recurrence and poor prognosis in patients who had primary colorectal cancer histologic grade with poorly differentiated or mucinous adenocarcinoma, over 3 num-bers of liver metastases, surgical margin of < 0.1 cm. Conclusions: This study showed that simultaneous resections present similar major complication rates and oncologic outcomes compared with staged resection for synchronous CRLM. Simultaneous resection for synchronous CRLM appears to be feasible and safe.

Matrix Metalloproteinase-1 Inhibitory Activity of Kaempferia pandurata Roxb.

Shim, Jae-Seok,Choi, Eun-Jung,Lee, Chan-Woo,Kim, Han-Sung,Hwang, Jae-Kwan The Korean Society of Food Science and Nutrition 2009 Journal of medicinal food Vol.12 No.3

Matrix metalloproteinase (MMP)-1 is a superfamily of zinc-dependent endopeptidases that are capable of degrading all components of the extracellular matrix. Kaempferia pandurata extract ($0.01-0.5\;{\mu}g/mL$) significantly reduced the expression of MMP-1 and induced the expression of type 1 procollagen at the protein and mRNA levels in a dosedependent manner. Ultraviolet (UV)-induced MMP-1 initiates cleavage of fibrillar collagen. Once cleaved by MMP-1, collagen can be further degraded by elevated levels of MMP-3 and MMP-9. It was found that increased MMP-1 expression due to UV irradiation was mediated by activation of mitogen-activated protein kinases such as extracellular-regulated kinase (ERK), Jun N-terminal kinase ( JNK), and p38 kinase. Treatment of K. pandurata extract in the range of $0.01-0.5\;{\mu}g/mL$ inhibited the UV-induced phosphorylations of ERK, JNK, and p38, respectively. Moreover, inhibition of phosphorylated ERK, JNK, and p38 by K. pandurata extract resulted in decreased c-Fos expression and c-Jun phosphorylation induced by UV light. The results strongly suggest that K. pandurata is potentially useful for the prevention and treatment of skin aging.

EPON 시스템 운용관리를 위한 구성관리블록 구현

심재찬(Jae Chan Shim),조경섭(Kyeong Seob Cho),김태일(Tae II Kim) 대한전자공학회 2003 대한전자공학회 학술대회 Vol.2003 No.11

이상치에 따른 불확실성에 관한 연구

심재원(Jae-Won Shim),남정찬(Jeong-Chan Nam),민희준(Hee-Jun Min),김가연(Ga-Yeon Kim),이효진(Hyo-Jin Lee),정혜영(Hye-Young Jeong) 한국지능시스템학회 2021 한국지능시스템학회논문지 Vol.31 No.4

딥러닝 모델의 성능을 평가하는 지표는 정확도를 평가할 뿐, 해당 모델의 불확실성을 평가할 수 없다. 불확실성은 정확도와는 별개로, 안정성 또는 신뢰도의 의미를 가질 수 있다. 불확실성을 측정할 수 있다면 기존 모델의 성능을 평가하는 지표 외에 새로운 지표가 될 가능성을 가지고 있다. 불확성을 측정하기 위한 딥러닝 모델로 베이지안 신경망을 사용한다. 본 논문은 불확실성이 딥러닝 모델에서 성능을 평가하는 지표가 될 수 있음을 확인해보려 하였다. 불확실성을 측정하기 위한 베이지안 신경망 모델로 베이지안 LSTM (Bayesian Long Short Term Memory)을 사용하였고, 모델의 학습에 실제 데이터를 적용하여 불확실성을 측정하였으며, 이를 기존 모델의 성능 지표와 비교분석 하였다. 실제 데이터를 이용한 실험에서 이상치가 추가됨에 따라 기존 모델의 성능지표는 증가하거나 감소하는 여러 가지 양상을 보이는 반면, 불확실성은 이상치에 따라 일관적으로 증가하는 결과를 보였다. Uncertainty is a separate measure of accuracy, and may have implications for stability or reliability. If uncertainty can be regulated, it has the potential to be a new metric for evaluating a performance of Deep learning models. Deep learning models for measuring uncertainty are Bayesian neural networks, which can be implemented by applying a drop-out scheme to each layer of existing deep learning models. This paper confirms that uncertainty can be an indicator of performance evaluation in deep learning models. We used Bayesian LSTM as a Bayesian neural network model to measure uncertainty, and measured uncertainty, which is compared to performance metrics from existing models.