• 전자공학회논문지B
• /
• 제31B권9호
• /
• pp.166-177
• /
• 1994

In this paper, an implemenation of on-line continuous recognizer for cursive Hangul handwriting is explained. For the Hangul recognition system, we propose a high speed string matching. The editing process in our proposed string matching is accomplished by single editing path. And the matching results are stored in a heap structure and we decide the user comfortibility of unceasing writing during recognition owing to the high speed matching. In the experimental result, a recongition rate of 86.36% at 1.75 second/character over 21,076 characters collected from 50 persons are abtained. And it is shown that the proposed recognition system is operated properly for the on-line recognition for cursive handwring between graphemes.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제3권2호
• /
• pp.178-186
• /
• 2003

Recently, many researchers have been interested in biometric systems such as fingerprint, handwriting, key-stroke patterns and human iris. From the viewpoint of reliability and robustness, iris recognition is the most attractive biometric system. Moreover, the iris recognition system is a comfortable biometric system, since the video image of an eye can be taken at a distance. In this paper, we discuss human iris recognition, which is based on accurate iris localization, robust feature extraction, and Neural Network classification. The iris region is accurately localized in the eye image using a multiresolution active snake model. For the feature representation, the localized iris image is decomposed using wavelet transform based on dyadic Haar wavelet. Experimental results show the usefulness of wavelet transform in comparison to conventional Gabor transform. In addition, we present a new method for setting initial weight vectors in competitive learning. The proposed initialization method yields better accuracy than the conventional method.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2000년도 제15차 학술회의논문집
• /
• pp.354-354
• /
• 2000

In this paper, we would study the applicability of neural networks to the recognition process of Korean stenographic character image, applying the classification function, which is the greatest merit of those of neural networks applied to the various pans so far, to the stenographic character recognition, relatively simple classification work. Korean stenographic recognition algorithms, which recognize the characters by using some methods, have a quantitative problem that despite the simplicity of the structure, a lot of basic characters are impossible to classify into a type. They also have qualitative one that it is not easy to classify characters for the delicacy of the character forms. Even though this is the result of experiment under the limited environment of the basic characters, this shows the possibility that the stenographic characters can be recognized effectively by neural network system. In this system, we got 90.86% recognition rate as an average.

• 로봇학회논문지
• /
• 제12권4호
• /
• pp.432-440
• /
• 2017

In this paper, we propose a jellyfish distribution recognition and monitoring system using a UAV (unmanned aerial vehicle). The UAV was designed to satisfy the requirements for flight in ocean environment. The target jellyfish, Aurelia aurita, is recognized through convolutional neural network and its distribution is calculated. The modified deep neural network architecture has been developed to have reliable recognition accuracy and fast operation speed. Recognition speed is about 400 times faster than GoogLeNet by using a lightweight network architecture. We also introduce the method for selecting candidates to be used as inputs to the proposed network. The recognition accuracy of the jellyfish is improved by removing the probability value of the meaningless class among the probability vectors of the evaluated input image and re-evaluating it by normalization. The jellyfish distribution is calculated based on the unit jellyfish image recognized. The distribution level is defined by using the novelty concept of the distribution map buffer.

• 조명전기설비학회논문지
• /
• 제12권3호
• /
• pp.74-82
• /
• 1998

본 연구에서는 초음파 센서에 의해 물체정보를 획득하고 불변모멘트 백터를 이용하여 이동 및 회전에 불변하는 물체특정점올 추출한다. 그리고 이를 SQFM(냉f요R없비1핑 Feature Map) 신경회로망의 입력데이터로 사용하여 물체의 이동 및 회전에 무관한 물체인식 시스템을 제안하였다. 또한 SOFM 신경회로망의 출력 neuron space 크기 및 반복학습회수와 물체인식률과의 관계를 실험하였다. 출력 neuron space와 반복학습회수를 각각 $4\times4~10\times10$까지, 10~50회까지 변화시쳐 물체인식올 실험한 결과 물체인식률은 동일한 값인 92.3[% 를 나타내었다.

• 대한전자공학회논문지
• /
• 제22권5호
• /
• pp.10-16
• /
• 1985

본 논문에서는 인쇄 양보 서상을 CCTV 카메라로써 마이크로 컴퓨터에 입력시켜, 이 화상을 인식, 스피커로 노래를 내어주는 컴퓨터 비젼 시스템에 관해 논하고 있다. 이때 내보서조의 특징추출 및 인식에는 가산투영법이 적용되구 그 대상 인식 범위는 내보의 여러 요소 중에서 오연 마디, 음표로 하고 있다. 아울러 분제 내보화징을 취급할 때 반드시 고려되어야 할 전처리 및 잡음 제거 과정을 보였고, 인식된 음표로 화음을 내민주는 간단한 하드웨어 시스템을 구성했다. 그 결과 보호한 인식률로 연주 가능함을 보였다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 ITC-CSCC -2
• /
• pp.785-788
• /
• 2000

The recognition system of a car plate is largely classified as the extraction and recognition of number plate. In this paper, we extract the number plate domain by using a thresholding method as a preprocess step. The computation of the density in a given mask provides a clue of a candidate domain whose density ratio corresponds to the properties of the number plate obtained in the best condition. The contour of the number plate for the recognition of the texts of number plate is extracted by operating Kohonen Algorithm in a localized region. The algorithm reduces noises around the contour. The recognition system with the density computation and Kohonen Algorithm shows a high performance in the real system in connection with a car number plate.

• 산업공학
• /
• 제13권2호
• /
• pp.188-194
• /
• 2000

A pattern recognition system is proposed to recognize characters in any type of Giro. The system consist of the character segmentation and the character recognition. Positional features from two round markers at the upper-right part and lower-left part of Giro is used for extracting character strings from images and RLE analysis is used if there are no round markers. A multi step combined method, which use a structural method and a statistical method, is used to improve recognition. The structural method apply rules on each characters, whereas a statistical method gives a different weighting vector to each pixel for improving the classification performance in regard to noises and distortions. The experimental results show that the proposed combined method has higher recognition rate, over than 98% even in cases that images are rotated about 10 degrees as well as have noises.

• 한국멀티미디어학회논문지
• /
• 제22권8호
• /
• pp.852-859
• /
• 2019

With the development of IT technology and smart devices, various applications utilizing image information are being developed. In order to provide an intuitive interface for pronunciation recognition, there is a growing need for research on pronunciation recognition using mouth feature values. In this paper, we propose a system to distinguish Korean vowel pronunciations by detecting feature points of lips region in images and applying Bayesian based learning model. The proposed system implements the recognition system based on Bayes' theorem, so that it is possible to improve the accuracy of speech recognition by accumulating input data regardless of whether it is speaker independent or dependent on small amount of learning data. Experimental results show that it is possible to effectively distinguish Korean vowels as a result of applying probability based Bayesian classification using only visual information such as mouth shape features.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권2호
• /
• pp.929-944
• /
• 2019

Many critical applications require accurate real-time human action recognition. However, there are many hurdles associated with capturing and pre-processing image data, calculating features, and classification because they consume significant resources for both storage and computation. To circumvent these hurdles, this paper presents a recognition machine learning (ML) based system model which uses reduced data structure features by projecting real 3D skeleton modality on virtual 2D space. The MMU VAAC dataset is used to test the proposed ML model. The results show a high accuracy rate of 97.88% which is only slightly lower than the accuracy when using the original 3D modality-based features but with a 75% reduction ratio from using RGB modality. These results motivate implementing the proposed recognition model on an embedded system platform in the future.