• 스마트미디어저널
• /
• 제5권1호
• /
• pp.38-43
• /
• 2016

스마트 환경 구축이 보편화됨에 따라 사람과 컴퓨터 사이의 상호작용(HCI)에 관한 연구가 활발히 진행되고 있다. 인간-컴퓨터 상호작용에서 사람의 얼굴과 시선 방향을 안다는 것은 그 사람의 의도나 관심의 대상을 파악하는데 중요한 정보를 제공할 뿐만 아니라 신체 구조를 이해하는데도 하나의 기준이 될 수 있으므로 중요한 연구 테마이다. 본 논문에서는 랜덤 포레스트를 이용하여 얼굴 방향을 미리 정해놓은 각도로 분류하는 방법을 제안한다. 먼저 영상은 전처리를 거친 뒤 회전정보를 얻기 위하여 평균 정면 얼굴과의 차영상을 이용하여 회전정보를 추출한다. 캐니에지 검출법을 이용하여 얼굴의 특징을 검출하고 이를 이용하여 에지 영상을 구한 뒤, 이 영상에 대해 가로 세로축 각각에 대해 픽셀 수를 누적하여 히스토그램을 작성한다. 누적히스토그램을 특징으로 랜덤 포레스트를 생성하였으며, 랜덤 포레스트의 학습과 테스트에는 CAS-PEAL-R1 데이터를 사용하여 80.6%의 인식률을 얻었다.

• Journal of Periodontal and Implant Science
• /
• 제26권3호
• /
• pp.735-751
• /
• 1996

An ultimate goal of periodontal therapy is to stop the disease process and to regenerate a functionally-oriented periodontium destroyed as a result of periodontal disease. The purpose of this study was to observe the effect of grafting granulation tissue obtained from extraction socket on the regeneration of horizontal furcation defect. Six dogs were used in this study. All mandibular first and third premolars were extracted. At 2, 3, and 5 days after extraction, tissues were obtained from extraction socket of 1 mongrel dog and examined by light microscope. Granulation tissue obtained at 5 days after extraction was chosen as the graft material. Five days later, horizontal furcation defects were created surgically at mandibular second and fourth premolars in the right and left side of the 5 beagle dogs. The entrance area of the artificially prepared "key hole" defects were about $3\;4mm^2$. By random selections, 2 exposed furcation defects were grafted with granulation tissue obtained from extraction socket as experimental group and 1 furcation defect was as control. The flaps were replaced to their original position and sutured with 4-0 chromic cat-gut. Three dogs were sacrificed 4 weeks and two dogs 8 weeks after surgery, and the prepared specimens were examined by light microscope. At 4 weeks, furcations were filled with epithelial lining and fibrous connective tissue infiltrated with chronic inflammatory cells. New bone formation was observed in all groups. Only experimental group showed new cementum formation. At 8 weeks, new cementum, functional arrangement of new PDL fiber, root resorption, and some ankylotic union of newly formed alveolar bone and root surface were observed in all groups. Experimental group showed that epithelial downgrowth was inhibited and new bone formation was more active compared to control. The success rate of the furcation defect healing was higher in experimental group than control. These results suggested that grafting of granulation tissue obtained from extraction socket which combined with reconstructive periodontal flap surgery may promote periodontal regeneration of horizontal furcation defect.

• 한국통신학회논문지
• /
• 제28권1C호
• /
• pp.24-31
• /
• 2003

In this paper, a new opto-digital multiple information hiding and real-time extracting system is implemented. That is, multiple information is hidden in a cover image by using the stego keys which are generated by combined use of random sequence(RS) and Hadamard matrix(HM) and these hidden information is extracted in real-time by using a new optical correlator-based extraction system. In the experiment, 3 kinds of information, English alphabet of "N", "R", "L" having 512$\times$512 pixels, are formulated 8$\times$8 blocks and each of these information is multiplied with the corresponding stego keys having 64$\times$64 pixels one by one. And then, by adding these modulated data to a cover image of "Lena"having 512$\times$512 pixels, a stego image is finally generated. In this paper, as an extraction system, a new optical nonlinear joint transform correlator(NJTC) is introduced to extract the hidden data from a stego image in real-time, in which optical correlation between the stego image and each of the stego keys is performed and from these correlation outputs the hidden data can be asily exacted in real-time. Especially, it is found that the SNRs of the correlation outputs in the proposed optical NJTC-based extraction system has been improved to 7㏈ on average by comparison with those of the conventional JTC system under the condition of having a nonlinear parameter less than k=0.4. This good experimental results might suggest a possibility of implementation of an opto-digital multiple information hiding and real-time extracting system.

• 셀메드
• /
• 제11권3호
• /
• pp.15.1-15.9
• /
• 2021

The chemical constitution of a drug has been accepted as an important basis for pharmacological action in Unani medicine. Various dosage forms have been developed on this concept, such as decoctions (Joshanda), infusions (Khesanda), extract (Rub / Usara), and syrup. Zarawand Mudaharaj (ZM.) / Aristolochia rotunda L. root was subjected to extraction process using Soxhlet's apparatus by using Response Surface Methodology (RSM) to design the number of random runs of the extracts with variation in the factors of temperature, the concentration of ethanol in water, time for extraction, for optimizing and maximizing the yield concentration. The data obtained, was analyzed with regression equation and ANOVA two-way summary to interpret the interaction of the factors for yield maximization. Minitab version 18 was used to design and analyze data. Validation of the optimum conditions for maximum yield of the whole extract of ZM. Root was carried out by re-run of the extract using the optimized conditions. The maximum yield percentage thus obtained using RSM was 20.87% whereas using these optimum conditions 21.35 % yield was obtained thereby validating the method. The association between the response functions and the process variables was identified by a three-factor recorded Box-Behnken design. In the present study RSM is used because itis a cheap and affordable method to optimize maximum yield percentage which may be reliably used by researchers. The study set in the surface conditions for ZM. root extraction by the Soxhlet apparatus for maximizing the yield percentage.

• ETRI Journal
• /
• 제42권1호
• /
• pp.90-100
• /
• 2020

Named entity recognition (NER) continues to be an important task in natural language processing because it is featured as a subtask and/or subproblem in information extraction and machine translation. In Urdu language processing, it is a very difficult task. This paper proposes various deep recurrent neural network (DRNN) learning models with word embedding. Experimental results demonstrate that they improve upon current state-of-the-art NER approaches for Urdu. The DRRN models evaluated include forward and bidirectional extensions of the long short-term memory and back propagation through time approaches. The proposed models consider both language-dependent features, such as part-of-speech tags, and language-independent features, such as the "context windows" of words. The effectiveness of the DRNN models with word embedding for NER in Urdu is demonstrated using three datasets. The results reveal that the proposed approach significantly outperforms previous conditional random field and artificial neural network approaches. The best f-measure values achieved on the three benchmark datasets using the proposed deep learning approaches are 81.1%, 79.94%, and 63.21%, respectively.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제11권3호
• /
• pp.1670-1683
• /
• 2017

Mitotic event recognition is a crucial and challenging task in biomedical applications. In this paper, we introduce the slow feature analysis and propose a fully-automated mitotic event recognition method for cell populations imaged with time-lapse phase contrast microscopy. The method includes three steps. First, a candidate sequence extraction method is utilized to exclude most of the sequences not containing mitosis. Next, slow feature is learned from the candidate sequences using slow feature analysis. Finally, a hidden conditional random field (HCRF) model is applied for the classification of the sequences. We use a supervised SFA learning strategy to learn the slow feature function because the strategy brings image content and discriminative information together to get a better encoding. Besides, the HCRF model is more suitable to describe the temporal structure of image sequences than nonsequential SVM approaches. In our experiment, the proposed recognition method achieved 0.93 area under curve (AUC) and 91% accuracy on a very challenging phase contrast microscopy dataset named C2C12.

• 인지과학
• /
• 제21권4호
• /
• pp.655-667
• /
• 2010

개체명 인식은 정보 추출의 한 단계로서 정보검색 분야 뿐 아니라 질의응답과 요약 분야에서 매우 유용하게 사용되고 있다. 본 논문에서는 structural Support Vector Machines(structural SVMs) 및 수정된 Pegasos 알고리즘을 이용한 한국어 개체명 인식 시스템에 대하여 기술하고 기존의 Conditional Random Fields(CRFs)를 이용한 시스템과의 성능을 비교한다. 실험결과 structural SVMs과 수정된 Pegasos 알고리즘이 기존의 CRFs 보다 높은 성능을 보였고(신뢰도 99%에서 통계적으로 유의함), structural SVMs과 수정된 Pegasos 알고리즘의 성능은 큰 차이가 없음(통계적으로 유의하지 않음)을 알 수 있었다. 특히 본 논문에서 제안하는 수정된 Pegasos 알고리즘을 이용한 경우 CRFs를 이용한 시스템보다 높은 성능(TV 도메인 F1=85.43, 스포츠 도메인 F1=86.79)을 유지하면서 학습 시간은 4%로 줄일 수 있었다.

• 오토저널
• /
• 제14권1호
• /
• pp.25-37
• /
• 1992

Interior noise in a car is known to have an important influence on product acceptability. This noise is closely correlated with structural-acoustic vibration. When considering noise problem, the structural-acoustic relation of a vehicle cabin model needs to be identified. However, it is very difficult to get the modal parameters of this kind of cabin structure composed of thin plates: because it not only can be excited by the acoustic vibration of cavity, but also tends to have decoupling effects of one plate from another. In order to obtain modal parameters more precisely, various excitation techniques, i.e. impact, pure random, burst random, and swept sine testing are applied for the first step. In the case of the cabin model, impact and swept sine testing show good results. Next, the determination of the excitation point by trial- and-error and the accurate measurements of FRF's are performed with these methods. The modal parameter extraction is carried out for the final step. This paper proposes a new approach to find the modal parameters more reliably in the case of high decoupling effects. That is, the convergence of MIF and MCF in each panel, which provide some criteria for the validity of the obtained modal parameters, is observed. And from those results, the pretty accurate modal parameters can be determined. A comparative assessment between the modal testing and the FEM is also performed.

• 한국정보처리학회논문지
• /
• 제3권6호
• /
• pp.1636-1645
• /
• 1996

본 논문에서는 패턴을 효과적으로 분류하기 위하여 화상자료의 특성인 이웃 화소 간의 종속성을 잘 표현해 주는 깁스확률장의 크리크를 바탕으로 2차원 조건부 모멘트 를 제안하였다. 이 알고리즘 구축은 공간정보를 갖는 조건부 모멘트를 이용하여 특정 벡터를 추출하는 과정과 패턴을 분류하는 과정으로 분리하여 생각한다. 특정벡터를 추출하는 과정은 하나의 패턴에 대해 깁스분포의 크리크로 표현된 파라미터를 추정한 다음, 2차원 조건부 모멘트들을 계산하여 특정벡터로부터 제안된 판별거리함수를 계 산하여 여러 원형 패턴 가운데 최소거리를 산출한 미지의 패턴을 원형패턴으로 분류 한다. 제안된 방법의 성능을 검증하기 위하여 대문자와 소문자 52자로 된 훈련 데이 타를 만들어 486 PC 66Mhz에서 실험을 한 결과 97.5% 이상의 분류성능이 있음을 밝혔 다.

• 전자공학회논문지C
• /
• 제36C권9호
• /
• pp.66-75
• /
• 1999

영상분할은 입력된 영상을 처리하여 유사한 화소들의 집합인 영역들로 화소들을 구분하는 작업이다. 영상분할의 결과는 영상인식의 정확성에 큰 영향을 미친다. 본 논문에서는 유전자 알고리즘을 이용하여 마르코프 랜덤 필드(Markov random field)에 기반한 영상분할 방법을 제안한다. 제안한 방법에서는 잡음과 흔들림(blurring)에 강한 MRF를 이용하여 영상을 모델링 한다. HRF기반 영상분할 방법은 왜곡에 강한 반면, 정확한 파라미터의 추정이 요구된다. 그래서 , 추정방법으로 많은 파라미터를 포함하는 문제를 다루는데 효율적인 유전자 알고리즘을 사용한다. 실 영상을 가지고 수행된 실험 결과와 자동 차량 추출 시스템에의 응용결과는 제안된 방법의 효율성을 보여준다.