• 제어로봇시스템학회논문지
• /
• 제4권2호
• /
• pp.246-255
• /
• 1998

This paper presents a novel method for real-time malfunction detection of plasma etching process using EPD signal traces. First, many reference EPD signal traces are collected using monochromator and data acquisition system in normal etching processes. Critical points are defined by applying differentiation and zero-crossing method to the collected reference signal traces. Critical parameters such as intensity, slope, time, peak, overshoot, etc., determined by critical points, and frame attributes transformed signal-to symbol of reference signal traces are saved. Also, UCL(Upper Control Limit) and LCL(Lower Control Limit) are obtained by mean and standard deviation of critical parameters. Then, test EPD signal traces are collected in the actual processes, and frame attributes and critical parameters are obtained using the above mentioned method. Process malfunctions are detected in real-time by applying SPC(Statistical Process Control) method to critical parameters. the Real-time malfunction detection method presented in this paper was applied to actual processes and the results indicated that it was proved to be able to supplement disadvantages of existing quality control check inspecting or testing random-selected devices and detect process malfunctions correctly in real-time.

• 대한의용생체공학회:의공학회지
• /
• 제17권1호
• /
• pp.109-120
• /
• 1996

Cardiac function is evaluated quantitatively using angiographic images via the analysis of the shape change or the heart wall boundaries. To kin with, boundary defection or ESLV(End Systolic Lert Ventricular) and EDLV(End Diastolic Left Ventricular) is essential for the quantitative analysis of cardiac function. The boundary detection methods proposed in the past were almost semi-automatic. Intervention by a knowledgeable human operator was still required Of con, manual tracing of the boundaries is currently used for subsequent analysis and diagnosis. This method would not cut excessive time, labor, and subjectivity associated with manual intervention by a human operator. EDLV images have noncontiguous and ambiguous edge signal on some boundary regions. In this paper, we propose a new method for automated detection of boundaries in noncontiguous and ambiguous EDLV images. The boundary detection scheme which based on a priori knowledge information is divided into two steps. The first step is to detect the candidate edge points of EDLV using ESLV boundaries. The second step is to correct detected boundaries of EDLV using the LV shape. We developed the algorithm of modifying EDLV boundaries defined adaptive modifier. We experimented the method proposed in this paper and compared our proposed method with the manual method in detecting boundaries of EDLV. In the areas within estimated boundaries of EDLV, the percentage of error was about 1.4%. We verified the useflilness and obtained the satisfying results througll the experiments of the proposed method.

• 대한의용생체공학회:의공학회지
• /
• 제44권5호
• /
• pp.346-353
• /
• 2023

In this paper, we propose an innovative approach that leverages deep learning to find optimal reference points for achieving precise tooth segmentation in three-dimensional tooth point cloud data. A dataset consisting of 350 aligned maxillary and mandibular cloud data was used as input, and both end coordinates of individual teeth were used as correct answers. A two-dimensional image was created by projecting the rendered point cloud data along the Z-axis, where an image of individual teeth was created using an object detection algorithm. The proposed algorithm is designed by adding various modules to the Unet model that allow effective learning of a narrow range, and detects both end points of the tooth using the generated tooth image. In the evaluation using DSC, Euclid distance, and MAE as indicators, we achieved superior performance compared to other Unet-based models. In future research, we will develop an algorithm to find the reference point of the point cloud by back-projecting the reference point detected in the image in three dimensions, and based on this, we will develop an algorithm to divide the teeth individually in the point cloud through image processing techniques.

• 한국식물병리학회지
• /
• 제14권6호
• /
• pp.630-635
• /
• 1998

Tobacco mosaic tobamovirus (TMV), cucumber mosaic cucumovirus (CMV), cucumber green mottle mosaic tobamovirus (CGMMV) and zucchini yellow mosaic potyvirus (ZYMV) from individual fruits and seeds of hot pepper and cucumber were detected by the reverse transcription-polymerase chain reaction (RT-PCR). The dilution end-points for RT-PCR in curde sap from TMV. and CMV - infected hot pepper leaves and CMV - and CGMMV-infected cucumber leaves were 10-5. However, the amount of PCR product obtained from preparation of ZYMV-infected cucumber leaf was 10-fold lower than those of CMV or CGMMV-infected cucumber leaves. In hot pepper, both TMV and CMV were detected in all parts of the fruit wall tissue, but the yields of PCR products in the fruit stalk and its surrounding tissues were higher than those of the end parts of the fruit. On the other hand, in cucumber fruit infected with CMV, CGMMV or ZYMV, the fruit wall tissue and seed located in both stalk and end parts showed higher yields of PCR products than those of intermediate parts. Of five viruses that were analysed, only TMV in hot pepper seed, and CGMMV and CMV in cucumber seed were detected in testa parts.

• Journal of Electrical Engineering and information Science
• /
• 제1권2호
• /
• pp.27-32
• /
• 1996

In target tracking area, the data association plays an important role and has been studied extensively. In this paper, after defining the data association as a constrained optimization, we introduce a new energy function and thereby an efficient realization of neural networks. As an application, this algorithm is used to detect object boundaries in IR images. The problem is that the IR image noisy, the shape of the object is variable, and the positions of the end points are not predictable. The performance of this algorithm is discussed with the experimental results.

• 한국정보전자통신기술학회논문지
• /
• 제9권2호
• /
• pp.229-236
• /
• 2016

본 논문에서는 3차원 공간에서 손 위치를 추적하고 움직임 궤적을 분석하여 원거리에서 판서가 가능한 기술을 제안한다. 3차원 공간에서 손으로 입력하는 한글 음절은 글자 획과 이동 획이 구분되지 않아 음절의 종류를 구분하기 힘들다. 이에 본 논문에서는 한글 음절을 구성하는 획을 글자 획과 이동 획으로 구분한 후 이동 획은 제거하고 글자 획만을 출력하는 방법을 제안한다. 우선, 필기체 음절의 궤적에서 획의 끝 점을 검출하고, 검출한 끝 점 정보를 이용하여 입력 음절을 획 단위로 분리한다. 음절 집합으로부터 8가지의 획 패턴을 정의한 후 분리한 획에 대해서는 방향 코드를 기반으로 획 패턴을 분류한다. 그리고 이를 기반으로 최종적으로 획의 유형을 글자 획과 이동 획으로 분류한다. 분류된 획의 유형을 기반으로 입력된 음절에서 이동 획은 제거하고 글자 획만을 출력하여 가독성이 있는 음절 표시가 가능하도록 한다. 360개의 음절 집합에 대해 정확도를 측정하여 획의 패턴은 88.3%, 획의 유형 구분은 91.1%의 정확도를 얻었다.

• 한국식물병리학회지
• /
• 제11권4호
• /
• pp.367-373
• /
• 1995

The cDNAs derived from the coat protein (CP) genes of six plant RNA viruses, tobacco mosaic virus-pepper strains (TMV-P) and -ordinary strain (TMV-OM), potato virus Y (PVY), turnip mosaic virus (TuMV), cucumber mosaic virus (CMV) and potato leafroll virus (PLRV), were subcloned into the transcription vector, pSPT18, containing SP6 and T7 promoters. The digoxigenin (DIG)-labeled RNA polymerase after linearlization of the cloned pSPTs with XbaI or SacI, and were tested for their sensitivities for the detection of the six viruses. In slot-blot hybridization, dilution end points for the detection of TMV-P and TMV-OM were 10-4, while those of PVY, TuMV and CMV were 10-3. PLRV was detected at the dilution of 10-2. When each RNA probe was applied for the detection of the viruses in the preparations from the leaf disks (8 mm in diameter, and 12 to 15 mg in weight) of infected natural host plants, TMV-P, TMV-OM and TuMV could be detected from one disk, while PVY from 1 or 2 disks. CMV was detected in the preparation from two disks, and PLRV from three disks. With DIG-labeled RNA probe, PVY was detected at 5 days after inoculation, but with ELISA the virus was detected at 8 days after inoculation to tobacco (Nicotiana tabacum cv. Xanthi nc) plants on which symptoms appeared at 9 days after inoculation. No difference was observed in cross reaction between the RNA probes for the detection of TMV-P and TMV-OM.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 2003년도 춘계학술대회 논문집
• /
• pp.992-995
• /
• 2003

A haptic device operated by the user's hand can receive information on position and orientation of the hand and display force and moment generated in the virtual environment to the hand. For realistic haptic display, the detailed information on collision between objects is necessary. In the past, the point-based graphic environment has been used in which the end effector of a haptic device was represented as a point and the interaction of this point with the virtual environment was investigated. In this paper, the shape-based graphic environment is proposed in which the interaction of the shape with the environment is considered to analyze collision or contact more accurately. To this end. the so-called Gilbert-Johnson-Keerthi (GJK) algorithm is adopted to compute collision points and collision instants between two shapes in the 3-D space. The 5- DOF haptic hand controller is used with the GJK algorithm to demonstrate a peg-in-hole operation in the virtual environment in conjunction with a haptic device. It is shown from various experiments that the shape-based representation with the GJK algorithm can provide more realistic haptic display for peg-in-hole operations.

• 한국컴퓨터정보학회논문지
• /
• 제10권6호
• /
• pp.1-8
• /
• 2005

얼굴 분석은 얼굴 인식 머리 움직임과 얼굴 표정을 이용한 인간과 컴퓨터사이의 인터페이스, 모델 기반 코딩, 가상현실 등 많은 응용 분야에서 유용하게 활용된다. 이러한 응용 분야에서는 얼굴의 특징점들을 정확하게 추출해야 한다. 본 논문에서는 눈, 눈썹, 입술의 코너와 같은 얼굴 특징을 자동으로 추출하는 방법을 제안한다. 먼저, 입력 영상으로부터 AdaBoost 기반의 객체 검출 기법을 이용하여 얼굴 영역을 추출한다. 그 다음에는 계곡 에너지. 명도 에너지, 경계선 에너지의 세 가지 특징 에너지를 계산하여 결합한다. 구해진 특징 에너지 영상에 대하여 에너지 값이 큰 수평 방향향의 사각형을 탐색함으로써 특징 영역을 검출한다. 마지막으로 특징 영역의 가장자리 부분에서 코너 검출 알고리즘을 적용함으로써 눈, 눈썹, 입술의 코너를 검출한다. 본 논문에서 제안된 얼굴 특징 추출 방법은 세 가지의 특징 에너지를 결합하여 사용하고 계곡 에너지와 명도 에너지의 계산이 조명 변화에 적응적인 특성을 갖도록 함으로써, 다양한 환경 조건하에서 견고하게 얼굴 특징을 추출할 수 있다.

• 한국멀티미디어학회논문지
• /
• 제24권5호
• /
• pp.659-666
• /
• 2021

With the continuous development of deep learning, human behavior recognition algorithms have achieved good results. However, in a multi-person recognition environment, the complex behavior environment poses a great challenge to the efficiency of recognition. To this end, this paper proposes a multi-person pose estimation model. First of all, the human detectors in the top-down framework mostly use the two-stage target detection model, which runs slow down. The single-stage YOLOv3 target detection model is used to effectively improve the running speed and the generalization of the model. Depth separable convolution, which further improves the speed of target detection and improves the model's ability to extract target proposed regions; Secondly, based on the feature pyramid network combined with context semantic information in the pose estimation model, the OHEM algorithm is used to solve difficult key point detection problems, and the accuracy of multi-person pose estimation is improved; Finally, the Euclidean distance is used to calculate the spatial distance between key points, to determine the similarity of postures in the frame, and to eliminate redundant postures.