• 한국정보과학회논문지:소프트웨어및응용
• /
• 제36권12호
• /
• pp.1016-1023
• /
• 2009

자동차 번호판 인식 시스템은 문자 추출, 특징 추출 등의 영상처리와 추출된 문자를 인식하는 인식기로 구성된다. 특징 추출은 문자 영역의 데이터 감소뿐만 아니라 인식 성능을 결정한다. 따라서 본 논문에서는 번호판 인식의 결과에 영향이 큰 숫자 인식, 특히 숫자의 특징 추출에 초점을 두었으며, 데이터의 군집성을 재배치하여 데이터 간의 최적의 산란도를 확보할 수 있는 통계적 특징의 혼합 모델을 제안하고, 이를 다층 퍼셉트론과 LVQ 신경망을 이용하여 유효성을 검증하였다. 제안된 통계적 특징 추출 방법은 번호판 영상이 갖는 정보를 가장 잘 유지하고, 잡음과 외부 환경에 강건하며 효과적인 방법임을 보여준다.

• 전자공학회논문지S
• /
• 제35S권8호
• /
• pp.96-105
• /
• 1998

본 논문에서는 화자 독립 연속 숫자음 인식 시스템의 성능향상을 위하여 MLP-VQ (Multi-Layer Perceptron-Vector Quantizer)를 이용한 가중 DHMM(WDHMM : Weighted Discrete Hidden Markov Models)을 제안한다. MLP 신경망의 출력분포는 입력 패턴과 학습 패턴들간의 비선형 매핑을 통해 각 패턴들간의 유사도를 나타내는 확률분포를 갖는다. 본 논문에서는 MLP 신경망의 출력분포중 가장 높은 출력 값을 갖는 MLP 신경망의 출력 노드를 인덱스를 이용하여 코드워드를 생성하는 MLP-VQ를 제안하였다. 제안된 MLP-VQ는 기존의 VQ에 비해 현재 입력패턴과 학습된 각 class 패턴들간의 유사성 정도를 인식모델을 반영할 수 있는 특징을 갖는다. 또한 MLP 신경망의 출력분포를 DHMM의 심벌 발생 확률의 가중치로 이용하는 가중 DHMM보다는 음소 클래스간의 관계를 인식모델에 반영할 수 있기 때문에 적은 계산양의 증가로 인식기의 성능을 14.71%개선할 수 있었다. 실험결과에 의하면, MLP-VQ와 WDHMM에 의한 화자독립 연결 숫자음 인식율은 84.22%이다.

• 정보보호학회논문지
• /
• 제30권4호
• /
• pp.527-536
• /
• 2020

정보보호용 디바이스의 부채널 정보인 소비 전력 파형을 이용하면 내장된 비밀 키 뿐만 아니라 동작 명령어를 복구할 수 있음이 밝혀졌다. 최근에는 MLP 등과 같은 딥러닝 모델을 이용한 프로파일링 기반의 부채널 공격들이 연구되고 있다. 본 논문에서는 마이크로 컨트롤러 AVR XMEGA128-D4가 사용하는 명령어에 대한 역어셈블러를 구현하였다. 명령어에 대한 템플릿 파형을 수집하고 전처리하는 과정을 자동화하였으며 CNN 딥러닝 모델을 사용하여 명령-코드를 분류하였다. 실험 결과, 전체 명령어는 약 87.5%의 정확도로, 사용 빈도가 높은 주요 명령어는 99.6%의 정확도로 분류될 수 있음을 확인하였다.

• 한국정보과학회논문지:컴퓨팅의 실제 및 레터
• /
• 제13권5호
• /
• pp.338-342
• /
• 2007

본 논문에서는 모바일 환경에서 고립단어 음성인식을 할 경우 화자종속 방법을 이용하여 성능을 높이는 사용자 적응형 후처리 방법을 제안한다. 이 방법은 인식기의 정확한 인식 결과를 위한 추가적인 처리들로 구성된다. 즉 인식기의 출력과 정확한 최종 결과들 간의 관계를 학습하여 이를 잘못된 인식기의 출력을 수정하는 데에 사용한다. 학습에는 패턴인식에 강인한 다층 퍼셉트론을 사용하며 학습 시간을 고려하여 모델을 세분화하고 동적으로 동작할 수 있도록 구현한다. 이 결과 인식기의 오류에 대해 41%를 수정하는 성과(오류 수정률: 41%)를 보였다.

• 대한의용생체공학회:학술대회논문집
• /
• 대한의용생체공학회 1990년도 춘계학술대회
• /
• pp.33-37
• /
• 1990

In this paper, a new method for lung area segmentation in chest radiographs has been presented. The movivation of this study is to include fuzzy informations about the relation between the image date structure and the area to be segmented in the segmentation process efficiently. The proposed method approached the segmentation problem in the perspective of pattern classification, using trainable pattern classifier, multi-layer perceptron. Having been trained with 10 samples, this method gives acceptable segmentation results, and also demonstrated the desirable property of giving better results as the training continues with more training samples.

• 제어로봇시스템학회논문지
• /
• 제13권4호
• /
• pp.371-376
• /
• 2007

This paper presents an emotion recognition and its expression system of an intelligent robot like a home robot or a service robot. Emotion recognition method in the robot is used by a facial image. We use a motion and a position of many facial features. apply a tracking algorithm to recognize a moving user in the mobile robot and eliminate a skin color of a hand and a background without a facial region by using the facial region detecting algorithm in objecting user image. After normalizer operations are the image enlarge or reduction by distance of the detecting facial region and the image revolution transformation by an angel of a face, the mobile robot can object the facial image of a fixing size. And materialize a multi feature selection algorithm to enable robot to recognize an emotion of user. In this paper, used a multi layer perceptron of Artificial Neural Network(ANN) as a pattern recognition art, and a Back Propagation(BP) algorithm as a learning algorithm. Emotion of user that robot recognized is expressed as a graphic LCD. At this time, change two coordinates as the number of times of emotion expressed in ANN, and change a parameter of facial elements(eyes, eyebrows, mouth) as the change of two coordinates. By materializing the system, expressed the complex emotion of human as the avatar of LCD.

• 전자공학회논문지C
• /
• 제35C권2호
• /
• pp.56-64
• /
• 1998

Multi-layered perceptrons that are a nonlinear neural network model, have been widely used for various applications mainly thanks to good function approximation capability for nonlinear fuctions. However, for digital hardware implementation of the multi-layere perceptrons, the quantization scheme using "look-up tables (LUTs)" is commonly employed to handle nonlinear signmoid activation functions in the neworks, and thus requires large amount of storage to prevent unacceptable quantization errors. This paper is concerned with a new effective methodology for digital hardware implementation of multi-layered perceptrons, and proposes a "piecewise affine approximation" method in which input domain is divided into (small number of) sub-intervals and nonlinear sigmoid function is linearly approximated within each sub-interval. Using the proposed method, we develop an expression and an error backpropagation type learning algorithm for a multi-layered perceptron, and compare the performance with the quantization method through Monte Carlo simulations on XOR problems. Simulation results show that, in terms of learning convergece, the proposed method with a small number of sub-intervals significantly outperforms the quantization method with a very large storage requirement. We expect from these results that the proposed method can be utilized in digital system implementation to significantly reduce the storage requirement, quantization error, and learning time of the quantization method.quantization method.

• 대한임베디드공학회논문지
• /
• 제11권3호
• /
• pp.163-171
• /
• 2016

Lane detection is a widely researched topic. Although simple road detection is easily achieved by previous methods, lane detection becomes very difficult in several complex cases involving noisy edges. To address this, we use a Convolution neural network (CNN) for image enhancement. CNN is a deep learning method that has been very successfully applied in object detection and recognition. In this paper, we introduce a robust lane detection method based on a CNN combined with random sample consensus (RANSAC) algorithm. Initially, we calculate edges in an image using a hat shaped kernel, then we detect lanes using the CNN combined with the RANSAC. In the training process of the CNN, input data consists of edge images and target data is images that have real white color lanes on an otherwise black background. The CNN structure consists of 8 layers with 3 convolutional layers, 2 subsampling layers and multi-layer perceptron (MLP) of 3 fully-connected layers. Convolutional and subsampling layers are hierarchically arranged to form a deep structure. Our proposed lane detection algorithm successfully eliminates noise lines and was found to perform better than other formal line detection algorithms such as RANSAC

• 제어로봇시스템학회논문지
• /
• 제21권10호
• /
• pp.958-964
• /
• 2015

In this paper, an object recognition method based on the depth information from the RGB-D camera, Xtion, is proposed for an indoor mobile robot. First, the RANdom SAmple Consensus (RANSAC) algorithm is applied to the point cloud obtained from the RGB-D camera to detect and remove the floor points. Next, the removed point cloud is classified by the k-means clustering method as each object's point cloud, and the normal vector of each point is obtained by using the k-d tree search. The obtained normal vectors are classified by the trained multi-layer perceptron as 18 classes and used as features for object recognition. To distinguish an object from another object, the similarity between them is measured by using Levenshtein distance. To verify the effectiveness and feasibility of the proposed object recognition method, the experiments are carried out with several similar boxes.

• 제어로봇시스템학회논문지
• /
• 제4권4호
• /
• pp.487-493
• /
• 1998

Camera calibration is the process of determining the coordinate relationship between a camera image and its real world space. Accurate calibration of a camera is necessary for the applications that involve quantitative measurement of camera images. However, if the camera plane is parallel or near parallel to the calibration board on which 2 dimensional objects are defined(this is called "ill-conditioned"), existing solution procedures are not well applied. In this paper, we propose a neural network-based approach to camera calibration for 2D images formed by a mono-camera or a pair of cameras. Multi-layer perceptrons are developed to transform the coordinates of each image point to the world coordinates. The validity of the approach is tested with data points which cover the whole 2D space concerned. Experimental results for both mono-camera and stereo-camera cases indicate that the proposed approach is comparable to Tsai's method［8］. Especially for the stereo camera case, the approach works better than the Tsai's method as the angle between the camera optical axis and the Z-axis increases. Therefore, we believe the approach could be an alternative solution procedure for the ill -conditioned camera calibration.libration.