• 한국통신학회논문지
• /
• 제21권11호
• /
• pp.3001-3011
• /
• 1996

In this paper, using DBNN algorithm which is used in the binary pattern classification or speech signal processing the digital DBNN circuit is designed having the variable expansion depending the size of input data and pattern type. The processing elemen(PE) of the proposed network consists of the synapse and MAXNET circuits for the similarity measurement between reference and input pattern. Global MAXNET selects the global winner among the local winners which is selected in each PE. Through the several simultions, and thus each PE and global MAXNET search the reference pattern that was the most simlar to input pattern for the discord of the pattern.

• 한국생산제조학회지
• /
• 제21권1호
• /
• pp.58-64
• /
• 2012

Instead of the surface pattern arranged repeatedly in two axial direction on a plane, we propose double patterns superposing two one-axial linear patterns as a reference target for surface encoding. A upper layer of the superposed pattern is the transparent glass with grooves cut in it at a fixed pitch. The position is sensed by detecting a shift of beam due to difference of a refractive index. And a lower layer is the aluminum with color-coated grooves. The amount of beam reflected on the layer varies according to its targeting position and is detected for encoding. For the above reference pattern, we can detect two-axial positions using only the single beam. Furthermore, the pattern size can be expanded with a size of the detector kept constant, meaning that the measured range can be expanded easily. In this paper, we review the existing optical encoding methods for grid pattern, and discuss the hardware implementation of the suggested surface encoding method.

• JSTS:Journal of Semiconductor Technology and Science
• /
• 제15권6호
• /
• pp.685-694
• /
• 2015

In this paper, a new dynamic reference scheme is proposed to improve the read voltage margin better than the previous static reference scheme. The proposed dynamic reference scheme can be helpful in compensating not only the background pattern dependence but also the cell position dependence. The proposed dynamic reference is verified by simulating the CMOS-memristor hybrid circuit using the practical CMOS SPICE and memristor Verilog-A models. In the simulation, the percentage read voltage margin is compared between the previous static reference scheme and the new dynamic reference scheme. Assuming that the critical percentage of read voltage margin is 5%, the memristor array size with the dynamic scheme can be larger by 60%, compared to the array size with the static one. In addition, for the array size of $64{\times}64$, the interconnect resistance in the array with the dynamic scheme can be increased by 30% than the static reference one. For the array size of $128{\times}128$, the interconnect resistance with the proposed scheme can be improved by 38% than the previous static one, allowing more margin on the variation of interconnect resistance.

• 한국멀티미디어학회논문지
• /
• 제18권3호
• /
• pp.312-322
• /
• 2015

In this paper a new defect detection method for flat panel display that does not require any separately prepared reference images and shows robustness against problems with regard to pixel tolerance and nonuniform illumination condition is proposed. In order to perform defect detection under any magnification value of camera, the proposed method automatically obtains the value of pattern interval through an image analysis. Using the information for pattern interval, an advanced PCSR-G method presented in this paper utilizes neighboring patterns as its reference images instead of utilizing any separately prepared reference images. Also this paper proposes a scheme to improve the performance of the conventional PCSR-G method by extracting and applying additional information for pixel tolerance and intensity distribution considering the value of pattern interval. Simulation results show that the performance of the proposed method utilizing pixel tolerance and intensity distribution is superior to that of the conventional method. Also, it is proved that the proposed method that is implemented using parallel technique based on GPGPU can be applied to real system.

• 대한방사선기술학회지:방사선기술과학
• /
• 제30권4호
• /
• pp.419-426
• /
• 2007

본 논문에서는 맛 인식을 위한 입력패턴벡터를 추출하고 패턴인식을 위한 맛(쓴맛, 단맛, 신맛, 짠맛)학습 알고리즘을 설계하였다. 입력패턴벡터의 구성을 위해 맛 활성화 신호의 세기가 사용되었고, 맛 패턴인식을 위한 알고리즘은 초기 참조벡터의 학습을 위해 SOM을 이용하였고, 종속 부류층의 출력뉴런의 부류지정을 위하여 out-star 학습법을 사용하였다. 제안된 알고리즘의 입력 층과 종속 클래스 층 사이의 연결강도는 SOM과 LVQ 알고리즘을 이용하여 초기 참조벡터의 설정 및 학습이 가능하게 하였다. 패턴벡터는 종속 부류층의 뉴런에 의해 종속 클래스로 분류하고, 종속 클래스 층과 출력 층 사이의 연결강도는 분류된 종속 부류를 클래스로 지정하는 학습을 하게 하였다. 패턴 분류를 위하여 제안된 학습알고리즘을 이용하여 시뮬레이션 되었고 기존의 LVQ 학습방식보다 우수한 분류성공률을 확인하였다.

• 한국컴퓨터정보학회논문지
• /
• 제11권5호
• /
• pp.95-103
• /
• 2006

본 논문에서는 뇌파인식을 위한 입력패턴벡터를 추출하고 패턴인식을 위한 뇌파 학습 알고리즘을 설계하였다. 입력패턴벡터의 구성을 위해 알파리듬과 베타리듬의 주파수와 진폭이 사용되었고, 뇌파패턴인식을 위한 알고리즘은 초기 참조벡터의 학습을 위해 SOM을 이용하고, 종속 부류층의 출력뉴런의 부류지정을 위하여 out-star 학습법을 사용하였다. 제안된 알고리즘의 입력 층과 종속 클래스 층 사이의 연결강도는 SOM과 LVQ 알고리즘을 이용하여 초기 참조벡터의 설정 및 학습이 가능하게 하였고, 패턴벡터를 종속 부류층의 뉴런에 의해 종속 클래스로 분류하고, 종속 클래스 층과 출력 층 사이의 연결강도는 분류된 종속 부류를 클래스로 지정하는 학습을 하게 된다. 뇌파 패턴 분류를 위하여 제안된 학습알고리즘을 이용하여 시뮬레이션 되었고 기존의 LVQ 학습방식보다 우수한 분류성공률을 확인하였다.

• 전자공학회논문지B
• /
• 제30B권10호
• /
• pp.95-104
• /
• 1993

IPA(interpattern association) model is a method of feature extraction using a neural network. Even in the case that the reference patterns are simuklar to one another, this model can recover the reference patterns effectively. However, when the pattern whose feature pixels are lost is used as input, this model can not guarantee perfect recovery of the reference pattern. It is proposed a improved interpattern association(IPA) model for the feature extraction using neural network. The improved IPA model that combines the first interconnection weight matrix of the IPA model with the second additional weight matrix is proposed here to overcome the recovery problem of the original IPA model. The results of computer simulation and optical experiment are advanced.

• 한국정밀공학회지
• /
• 제21권7호
• /
• pp.83-91
• /
• 2004

The determination of camera position and orientation from known correspondences of 3D reference points and their images is known as pose estimation in computer vision or space resection in photogrammetry. This paper discusses estimation of transform parameters using the pattern matching method with 2D images only. In general, the 3D reference points or lines are needed to find out the 3D transform parameters, but this method is applied without the 3D reference points or lines. It uses only two images to find out the transform parameters between two image. The algorithm is simulated using Visual C＋＋ on Windows 98.

• Journal of information and communication convergence engineering
• /
• 제8권3호
• /
• pp.339-342
• /
• 2010

An iris image pattern positioning with preserved edge detector, ring zone and clock zone, frequency distribution and overlay matching is presented in this paper. Edge detector is required to be powerful and detail. That is proposed by overlaying Canny with LOG (CLOG). The two reference patterns are made from allocating each gray level on the clock zone and ring zone respectively. The normalized target image is overlaid with the clock zone reference pattern and the ring zone pattern to extract overlapped number, and make a matched frequency distribution to look through a symptom and position of human organ and tissue. The iterating experiments result in the ring and clock zone positioning evaluation.

• 한국정보통신학회논문지
• /
• 제20권10호
• /
• pp.1973-1979
• /
• 2016

기술이 발전하면서 카메라를 통해 3D 측정을 하는 방법은 계속 발전되어왔고 최근에는 여러 주기의 fringe pattern을 이용한 측정 방법을 쓰고 있다. 본 논문에서는 여러 주기의 fringe pattern을 이용한 3D 측정 방법에 대한 기존 방법의 문제점을 제시하고, 이에 대한 해결 방안으로 기준면의 절대위상과 물체의 위상차를 이용한 물체의 절대위상을 구하는 방법을 제안한다. 이를 이용하면 새로운 물체에 대해서 매번 여러 주기의 fringe pattern을 조사하지 않고 물체의 절대위상을 얻을 수 있다. 따라서 제안하는 방법을 이용하면, 측정단계에서 취득하는 영상의 개수가 적기 때문에 보다 빠른 속도로 3D 측정을 할 수 있다. 실험을 통하여 제안하는 방법의 유용성을 보였다.