• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.10
• /
• pp.1341-1347
• /
• 2020

Legacy studies for classifying arrhythmia have been studied in order to improve the accuracy of classification, Neural Network, Fuzzy, Machine Learning, etc. In particular, deep learning is most frequently used for arrhythmia classification using error backpropagation algorithm by solving the limit of hidden layer number, which is a problem of neural network. In order to apply a deep learning model to an ECG signal, it is necessary to select an optimal model and parameters. In this paper, we propose parameter extraction based on AR and arrhythmia classification through a deep learning. For this purpose, the R-wave is detected in the ECG signal from which noise has been removed, QRS and RR interval is modelled. And then, the weights were learned by supervised learning method through deep learning and the model was evaluated by the verification data. The classification rate of PVC is evaluated through MIT-BIH arrhythmia database. The achieved scores indicate arrhythmia classification rate of over 97%.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.7
• /
• pp.898-903
• /
• 2020

In this paper, we propose an indoor logistic tracking system that identifies the location and inventory of the logistics in the room based on fingerprints. Through this, we constructed the actual infrastructure of the logistics center and designed and implemented the logistics management system. The proposed system collects the signal strength through the location terminal and generates the signal map to locate the goods. The location terminal is composed of a UHF RFID reader and a wireless LAN card, reads the peripheral RFID signal and the signal of the wireless AP, and transmits it to the web server. The web server processes the signal received from the location terminal and stores it in the database, and the user uses the data to produce the signal map. The proposed system combines UHF RFID with existing fingerprinting method to improve performance in the environment of querying multiple objects.

• Journal of Korea Society of Industrial Information Systems
• /
• v.25 no.6
• /
• pp.15-23
• /
• 2020

While research activities in the agricultural field for climate change are being actively carried out, smart agriculture using information and communication technology has become a new trend in line with the Fourth Industrial Revolution. Accordingly, research is being conducted to identify and respond to signs of abnormal growth in advance by monitoring the stress of crops in various outdoor environments and soil conditions. There are also attempts to analyze data collected in real time through various sensors using artificial intelligence techniques or big data technologies. In this paper, we propose a big data model that is effective in analyzing the growth environment informations and biometric information of crops by using the existing relational database for big data analysis. The performance of the model was measured by the response time to a query according to the amount of data. As a result, it was confirmed that there is a maximum time reduction effect of 23.8%.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.17 no.2
• /
• pp.29-40
• /
• 2007

The purpose of this research is to propose a method for scoring the quality of a fingerprint image using the local information derived from the fingerprint image. In previous works for the quality measuring, most of the quality scores are related to the performance of a matching algorithm, and this makes the quality result more subjective. The quality score of a fingerprint image proposed in this work is sensor-independent, source-independent and matcher-independent one, and this concept of fingerprint sample quality results in effective improvement of the system performance. In this research, a new definition of fingerprint image quality and a new method for measuring the quality are proposed. For the experiments, several sub-databases from FVCs are used and the proposed method showed reasonable results for the test database. The proposed method can be used in various systems for the numerous purposes since the quality scores generated by the proposed method are based on the idea that the quality of fingerprint should be sensor-independent, source-independent and matcher-independent.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.21 no.1
• /
• pp.85-91
• /
• 2011

The technique, which is able to detect and recognize characters in videos such as a movie or TV drama, can be used for applications which are database management of a general user's facial images for the suppliers of PVR(personal video recorder), mobile phones, and multimedia, etc. In this paper, we propose a face detection algorithm. It searches the character through cast indexing when the scene is changed in video. It is consisted of three stages. The first step is the detection-step of the scene change after producing a paused image. The second step is the face detection-step using color information. The final step is the detection-step which detects its features by the facial boundary. According to the experimental result, it has detected faces in different conditions successfully and more advanced than the existing other one that are using only color information.

• The Journal of Korean Institute of Information Technology
• /
• v.17 no.8
• /
• pp.11-19
• /
• 2019

Against advanced image-related crimes, a high level of digital forensic methods is required. However, feature-based methods are difficult to respond to new device features by utilizing human-designed features, and deep learning-based methods should improve accuracy. This paper proposes a deep learning model to identify camera models based on DenseNet, the recent technology in the deep learning model field. To extract camera sensor features, a HPF feature extraction filter was applied. For camera model identification, we modified the number of hierarchical iterations and eliminated the Bottleneck layer and compression processing used to reduce computation. The proposed model was analyzed using the Dresden database and achieved an accuracy of 99.65% for 14 camera models. We achieved higher accuracy than previous studies and overcome their disadvantages with low accuracy for the same manufacturer.

• Phonetics and Speech Sciences
• /
• v.12 no.4
• /
• pp.63-71
• /
• 2020

Speech recognition is one of the areas actively commercialized using deep learning and machine learning techniques. However, the majority of speech recognition systems on the market are developed on data with limited diversity of speakers and tend to perform well on typical adult speakers only. This is because most of the speech recognition models are generally learned using a speech database obtained from adult males and females. This tends to cause problems in recognizing the speech of the elderly, children and people with dialects well. To solve these problems, it may be necessary to retain big database or to collect a data for applying a speaker adaptation. However, this paper proposes that a new end-to-end speech recognition method consists of an acoustic augmented recurrent encoder and a transformer decoder with linguistic prediction. The proposed method can bring about the reliable performance of acoustic and language models in limited data conditions. The proposed method was evaluated to recognize Korean elderly and children speech with limited amount of training data and showed the better performance compared of a conventional method.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.25 no.1
• /
• pp.103-111
• /
• 2021

Naju Seokdanggan, Treasure No. 49, was dismantled and reconstructed due to poor performance. During construction, the crack area was reinforced and the inclination was improved. It is necessary to analyze the stiffness changes before and after the reconstruction of these cultural properties, and to establish a database of related information. In addition, there is a need for research on a scientific non-destructive testing method capable of predicting or evaluating the reinforcing effect. In this study, a simple equation for estimating the overall stiffness of the structural system was derived from information on the elasticity coefficient and the natural frequency measured by vibration tests before and after reconstruction work, and the applicability of the equation was examined. If the stiffness of important cultural properties is regularly investigated by the suggested method, it is judged that it can be used as data to estimate the time when structural safety diagnosis is necessary or when repair or reinforcement is necessary.

• Journal of Convergence for Information Technology
• /
• v.12 no.4
• /
• pp.126-132
• /
• 2022

At present, due to the rapid spread of smartphones and activation of IoT, malicious codes are disseminated using SNS, or intelligent intrusions such as intelligent APT and ransomware are in progress. The damage caused by the intelligent intrusion is also becoming more consequential, threatening, and emergent than the previous intrusion. Therefore, in this paper, we propose an intelligent intrusion situation-aware reasoning system to detect transgression behavior made by such intelligent malicious code. The proposed system was used to detect and respond to various intelligent intrusions at an early stage. The anticipated system is composed of an event monitor, event manager, situation manager, response manager, and database, and through close interaction between each component, it identifies the previously recognized intrusive behavior and learns about the new invasive activities. It was detected through the function to improve the performance of the inference device. In addition, it was found that the proposed system detects and responds to intelligent intrusions through the state of detecting ransomware, which is an intelligent intrusion type.

• Journal of Korea Society of Industrial Information Systems
• /
• v.27 no.2
• /
• pp.1-10
• /
• 2022

Deception technology is an extended concept of honeypot, which detects, prevents or delays attacks by deceiving adversaries. It has been applied to various system components such as network ports, services, processes, system calls and database management systems. We can apply the same concept to attacks on files. A representative example of a file attack is ransomware. Ransomware is a type of malware that encrypts user files and ask for ransom to recover those files. Another example is the wiper attack, which erases all or target files of a system. In this paper we propose a defense mechanism against these kinds of attacks by hiding files. Compared to backup or virtualization techniques, the proposed method incurs less space and performance overheads.