• The Journal of the Acoustical Society of Korea
• /
• v.30 no.3
• /
• pp.149-157
• /
• 2011

In this paper, we propose a new cardiac disorder classification method using an support vector machine (SVM) to combine hidden Markov model (HMM) and murmur existence information. Using cepstral features and the HMM Viterbi algorithm, we segment input heart sound signals into HMM states for each cardiac disorder model and compute log-likelihood (score) for every state in the model. To exploit the temporal position characteristics of murmur signals, we divide the input signals into two subbands and compute murmur probability of every subband of each frame, and obtain the murmur score for each state by using the state segmentation information obtained from the Viterbi algorithm. With an input vector containing the HMM state scores and the murmur scores for all cardiac disorder models, SVM finally decides the cardiac disorder category. In cardiac disorder classification experimental results, the proposed method shows the relatively improvement rate of 20.4 % compared to the HMM-based classifier with the conventional cepstral features.

• KIISE Transactions on Computing Practices
• /
• v.23 no.5
• /
• pp.316-321
• /
• 2017

We analyze the online Hangul handwriting recognition problem (HHR) and present solutions based on recurrent neural networks. The solutions are organized according to the three kinds of sequence labeling problem - sequence classifications, segment classification, and temporal classification, with additional consideration of the structural constitution of Hangul characters. We present a stacked gated recurrent unit (GRU) based model as the natural HHR solution in the sequence classification level. The proposed model shows 86.2% accuracy for recognizing 2350 Hangul characters and 98.2% accuracy for recognizing the six types of Hangul characters. We show that the type recognizing model successfully follows the type change as strokes are sequentially written. These results show the potential for RNN models to learn high-level structural information from sequential data.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.12
• /
• pp.587-594
• /
• 2021

As the diagnosis using encephalography(EEG) has been expanded, various studies have been actively performed for classifying EEG automatically. This paper proposes a CNN model that can effectively classify EEG signals acquired from healthy persons and patients with epilepsy. We segment the EEG signals into sub-signals with smaller dimension to augment the EEG data that is necessary to train the CNN model. Then the sub-signals are segmented again with overlap and they are used for training the CNN model. We also propose ensemble strategy in order to improve the classification accuracy. Experimental result using public Bonn dataset shows that the CNN can detect the epileptic seizure with the accuracy above 99.0%. It also shows that the ensemble method improves the accuracy of 3-class and 5-class EEG classification.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.16 no.3
• /
• pp.232-247
• /
• 1991

This paper proposes a new classification algorithm using characteristic and structural information of printed Hanja as preliminary stages of Hanja-character recognition. Hanja is difficult for not only recognition but classification as many character and complicated structure. In this paper, to solve thie problem, extracted common subpattern in classified pattern after processing type classification fot Hanja pattern. First, we extracted subpattern, after we process preprecessing about input of character pattern, extracting directional segment, labeling on 4-directional pattern and 12 type classified using structural information based on the subpattern existing region of character pattern. Though the experiment, this study obtained that classified rate of Hanja is 93.07% on 1800 character of educational Hanja and 90.12% on 4888 character of KS C5601 standard TRIGEM LBP Hanja font and saw that as extracting subpattern at classified data was this paper possibly applied to the recognition.

• Journal of the Semiconductor & Display Technology
• /
• v.10 no.2
• /
• pp.31-38
• /
• 2011

This paper introduces a video based traffic monitoring system for detecting vehicles and obstacles on the road. To segment moving objects from image sequence, we adopt the background subtraction algorithm based on the local binary patterns (LBP). Recently, LBP based texture analysis techniques are becoming popular tools for various machine vision applications such as face recognition, object classification and so on. In this paper, we adopt an extension of LBP, called the Diagonal LBP (DLBP), to handle the background subtraction problem arise in vision-based autonomous parking systems. It reduces the code length of LBP by half and improves the computation complexity drastically. An edge based shadow removal and blob merging procedure are also applied to the foreground blobs, and a pose estimation technique is utilized for calculating the position and heading angle of the moving object precisely. Experimental results revealed that our system works well for real-time vehicle localization and tracking applications.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.1
• /
• pp.24-31
• /
• 2020

In this paper, the deep convolutional neural network for sound event detection is studied. Especially, the end-to-end neural network, which generates the detection results from the input audio waveform, is studied for weakly supervised problem that includes weakly-labeled and unlabeled dataset. The proposed system is based on the network structure that consists of deeply-stacked 1-dimensional convolutional neural networks, and enhanced by the skip connection and gating mechanism. Additionally, the proposed system is enhanced by the sound event detection and post processings, and the training step using the mean-teacher model is added to deal with the weakly supervised data. The proposed system was evaluated by the Detection and Classification of Acoustic Scenes and Events (DCASE) 2019 Task 4 dataset, and the result shows that the proposed system has F₁-scores of 54 % (segment-based) and 32 % (event-based).

• 제어로봇시스템학회:학술대회논문집
• /
• 1988.10a
• /
• pp.697-700
• /
• 1988

This paper describes the design of a hardware system for ECG feature extraction based on pipeline processor consisting of three computers. ECG data is acquisited by 12 bit A/D converter with hardware QRS triggred detector. Four diagnostic parameters-heart, axis, and ST axis, and ST segment are used for the classification and the diagnosis of arrhythmia. The functions of the main CPU were distributed and processed with three microcomputers. Therefore the effective data process and the real time process using microcomputer can be obtained. The interconnection structure consisting of two common memory units is designed to decrease the delay time caused by data transfer between processors and designed by which the delay time can be taken 1% of one clock period.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.26 no.8
• /
• pp.1217-1226
• /
• 1989

This paper describes a design of hardware system for real time automatic diagnosis of ECG arrhythmia based on pipeline processor consisting of three microcomputer. ECG data is acquisited by 12 bit A/D converter with hardware QRS triggered detector. Four diagnostic parameters-heart rate, morpholigy, axis, and ST segment-are used for the classification and the diagnosis of arrhythmia. The functions of the main CPU were distributed and processed with three microcomputers. Therefore the effective data process and the real time process using microcomputer can be obtained. The interconnection structure consisting of two common memory unit is designed to decrease the delay time caused by data transfer between processors and be which the delay time can be taken 1% of one clock period.

• Proceedings of the KSRS Conference
• /
• 2009.03a
• /
• pp.207-212
• /
• 2009

고해상도 위성영상이 갖는 공간 객체의 복잡성과 다양성에 의해 기존 중 저해상도 영상에서 사용하던 분류 방식을 고해상도 영상에 그대로 적용하기에는 한계가 있다. 이러한 문제를 극복하기 위하여 영상은 공간적인 특성을 추가적으로 추출하여 분광정보와 결합하여 분류를 수행하는 방식의 연구가 진행되고 있다. 본 연구의 목적은 고해상도 영상의 분류정확도를 개선하기 위하여 새로운 공간 개체(spatial feature)인 SSI(Shape-Size Index)를 제안하는데 있다. SSI는 영역 확장(Region Growing) 기반의 영상 분할(Image Segmentation)을 수행한 후, 객체 내에 객체의 크기와 모양에 대한 고려를 모두 할 수 있는 공간 속성값을 할당하여 공간정보를 추출한다. 추출된 공간정보를 고해강도 영상의 다중분광 밴드와 결합하여 Support Vector Machine(SVM)을 이용한 분류를 수행하였다. 실험 결과, 제안한 기법의 분류 결과가 분광밴드만을 이용하여 분류를 수행한 결과뿐만 아니라 기존의 공간 개체 추출방식인 GLCM, PSI 기법을 이용한 분류 결과에 비해 높은 분류정확도를 도출함을 알 수 있었다.

• Proceedings of the KSRS Conference
• /
• 2008.10a
• /
• pp.200-203
• /
• 2008

The objective of this study is to produce the $CO_2$ (carbon dioxide) absorption map using KOMPSAT-2 imagery. For estimating the amount of $CO_2$ absorption, the stand biomass of forest was estimated with the total weight, which was the sum of individual tree weight. Individual tree volumes could be estimated by the crown width extracted from KOMPSAT-2 imagery. In particular, the carbon conversion index and the ratio of the $CO_2$ molecular weight to the C atomic weight, reported in the IPCC (Intergovernmental Panel on Climate Change) guideline, was used to convert the stand biomass into the amount of $CO_2$ absorption. Thereafter, the KOMPSAT-2 imagery was classified with the SBC (segment based classification) method in order to quantify $CO_2$ absorption by tree species. As a result, the map of $CO_2$ absorption was produced and the amount of $CO_2$ absorption was estimated by tree species.