• Journal of the Korea Society of Computer and Information
• /
• v.25 no.10
• /
• pp.35-42
• /
• 2020

In this study, we propose a crack detection method using limited data with a U-Net based image inpainting technique that is a modified unsupervised anomaly detection method. Concrete cracking occurs due to a variety of causes and is a factor that can cause serious damage to the structure in the long term. In general, crack investigation uses an inspector's visual inspection on the concrete surfaces, which is less objective in judgment and has a high possibility of human error. Therefore, a method with objective and accurate image analysis processing is required. In recent years, the methods using deep learning have been studied to detect cracks quickly and accurately. However, when the amount of crack data on the building or infrastructure to be inspected is small, existing crack detection models using it often show a limited performance. Therefore, in this study, an unsupervised anomaly detection method was used to augment the data on the object to be inspected, and as a result of learning using the data, we confirmed the performance of 98.78% of accuracy and 82.67% of harmonic average (F1_Score).

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2003.05a
• /
• pp.155-160
• /
• 2003

This paper presents a multiple face detector based on a robust pupil detection technique. The pupil detector uses active illumination that exploits the retro-reflectivity property of eyes to facilitate detection. The detection range of this method is appropriate for interactive desktop and kiosk applications. Once the location of the pupil candidates are computed, the candidates are filtered and grouped into pairs that correspond to faces using heuristic rules. To demonstrate the robustness of the face detection technique, a dual mode face tracker was developed, which is initialized with the most salient detected face. Recursive estimators are used to guarantee the stability of the process and combine the measurements from the multi-face detector and a feature correlation tracker. The estimated position of the face is used to control a pan-tilt servo mechanism in real-time, that moves the camera to keep the tracked face always centered in the image.

• Journal of the Korea Society of Computer and Information
• /
• v.17 no.11
• /
• pp.73-81
• /
• 2012

This paper proposes an ETWAD(Encapsulation and Tunneling Wormhole Attack Detection) design based on positional information and hop count on Ad-Hoc Network. The ETWAD technique is designed for generating GAK(Group Authentication Key) to ascertain the node ID and group key within Ad-hoc Network and authenticating a member of Ad-hoc Network by appending it to RREQ and RREP. In addition, A GeoWAD algorithm detecting Encapsulation and Tunneling Wormhole Attack by using a hop count about the number of Hops within RREP message and a critical value about the distance between a source node S and a destination node D is also presented in ETWAD technique. Therefore, as this paper is estimated as the average probability of Wormhole Attack detection 91%and average FPR 4.4%, it improves the reliability and probability of Wormhole Attack Detection.

• Journal of KIISE:Information Networking
• /
• v.32 no.3
• /
• pp.301-309
• /
• 2005

Learning program's behavior using machine learning techniques based on system call audit data is an effective intrusion detection method. Rule teaming, neural network, statistical technique, and hidden Markov model are representative methods for intrusion detection. Among them neural networks are known for its good performance in teaming system call sequences. In order to apply it to real world problems successfully, it is important to determine their structure. However, finding appropriate structure requires very long time because there are no formal solutions for determining the structure of networks. In this paper, a novel intrusion detection technique using evolutionary neural networks is proposed. Evolutionary neural networks have the advantage that superior neural networks can be obtained in shorter time than the conventional neural networks because it leams the structure and weights of neural network simultaneously Experimental results against 1999 DARPA IDEVAL data confirm that evolutionary neural networks are effective for intrusion detection.

• Journal of the Korea Society of Computer and Information
• /
• v.17 no.9
• /
• pp.49-56
• /
• 2012

In this paper, we propose the method that decides the shot types using face detection technique. The shot types, such as close-up shot, medium shot, and long shot, can be applied as useful information for understanding narrative structure of movies. The narrative structure of movie is builded by characters. Also their mental and emotional changes become inextricably bound up with them of narrative. The shot types are decided by distance between character and camera. If put together above them, shot types can be found by using detection technique of face size of characters and understand narrative of movie. To do this, we propose the methodology to detect shot type by face detecting and implement the system to do it. Additionally, we evaluate the performance of the system. The implementation system has been evaluated as 95% for close-up shot detection and 90% for medium shot detection, while 53.3% is just detected for long shots.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.7 no.8
• /
• pp.1823-1830
• /
• 2003

This paper presents a multiple face detector based on a robust pupil detection technique. The pupil detector uses active illumination that exploits the retro­reflectivity property of eyes to facilitate detection. The detection range of this method is appropriate for interactive desktop and kiosk applications. Once the location of the pupil candidates are computed, the candidates are filtered and grouped into pairs that correspond to faces using heuristic rules. To demonstrate the robustness of the face detection technique, a dual mode face tracker was developed, which is initialized with the most salient detected face. Recursive estimators are used to guarantee the stability of the process and combine the measurements from the multi­face detector and a feature correlation tracker. The estimated position of the face is used to control a pan­tilt servo mechanism in real­time, that moves the camera to keep the tracked face always centered in the image.

• Smart Structures and Systems
• /
• v.30 no.6
• /
• pp.613-626
• /
• 2022

Guaranteeing the quality and integrity of structural health monitoring (SHM) data is very important for an effective assessment of structural condition. However, sensory system may malfunction due to sensor fault or harsh operational environment, resulting in multiple types of data anomaly existing in the measured data. Efficiently and automatically identifying anomalies from the vast amounts of measured data is significant for assessing the structural conditions and early warning for structural failure in SHM. The major challenges of current automated data anomaly detection methods are the imbalance of dataset categories. In terms of the feature of actual anomalous data, this paper proposes a data anomaly detection method based on data-level and deep learning technique for SHM of civil engineering structures. The proposed method consists of a data balancing phase to prepare a comprehensive training dataset based on data-level technique, and an anomaly detection phase based on a sophisticatedly designed network. The advanced densely connected convolutional network (DenseNet) and Transformer encoder are embedded in the specific network to facilitate extraction of both detail and global features of response data, and to establish the mapping between the highest level of abstractive features and data anomaly class. Numerical studies on a steel frame model are conducted to evaluate the performance and noise immunity of using the proposed network for data anomaly detection. The applicability of the proposed method for data anomaly classification is validated with the measured data of a practical supertall structure. The proposed method presents a remarkable performance on data anomaly detection, which reaches a 95.7% overall accuracy with practical engineering structural monitoring data, which demonstrates the effectiveness of data balancing and the robust classification capability of the proposed network.

• The Journal of the Acoustical Society of Korea
• /
• v.31 no.7
• /
• pp.449-460
• /
• 2012

The importance of the leak detection of a buried pipe in a power plant of Korea is being emphasized as the buried pipes of a power plant are more than 20 years old. The objective of this work is to enhance the capability of the leak detection technique in a noisy environment. For this purpose, a modified cross-correlation method that can effectively remove the rotating machinery noise component is suggested. In addition, a method for leak point detection using phase information of cross-spectrum is suggested. The validity of the proposed method is verified by performing an experiment. The experimental result demonstrates that the performance of the cross-correlation method can be enhanced by reducing the periodic noise components due to mechanical equipment.

• Smart Structures and Systems
• /
• v.20 no.4
• /
• pp.473-481
• /
• 2017

This study suggests a simple, convenient and non-destructive method for investigation of the Young's modulus detection in stepped shafts which only utilizes the first-order resonant frequency in flexural mode and dimensions of structures. The method is based on the impulse excitation technique (IET) to pick up the fundamental resonant frequencies. The standard Young's modulus detection formulas for rectangular and circular cross-sections are well investigated in literatures. However, the Young's modulus of stepped shafts can not be directly detected using the formula for a beam with rectangular or circular cross-section. A response surface method (RSM) is introduced to design numerical simulation experiments to build up experimental formula to detect Young's modulus of stepped shafts. The numerical simulation performed by finite element method (FEM) to obtain enough simulation data for RSM analysis. After analysis and calculation, the relationship of flexural resonant frequencies, dimensions of stepped shafts and Young's modulus is obtained. Numerical simulations and experimental investigations show that the IET method can be used to investigate Young's modulus in stepped shafts, and the FEM simulation and RSM based IET formula proposed in this paper is applicable to calculate the Young's modulus in stepped shaft. The method can be further developed to detect mechanical parameters of more complicated structures using the combination of FEM simulation and RSM.

• Smart Structures and Systems
• /
• v.16 no.6
• /
• pp.1107-1132
• /
• 2015

A damage detection algorithm based on neuro fuzzy hybrid system is presented in this study for location and severity predictions of cracks in beam-like structures. A combination of eigenfrequencies and rotation deviation curves are utilized as input to the soft computing technique. Both single and multiple damage cases are considered. Theoretical expressions leading to modal properties of damaged beam elements are provided. The beam formulation is based on Euler-Bernoulli theory. The cracked section of beam is simulated employing discrete spring model whose compliance is computed from stress intensity factors of fracture mechanics. A hybrid neuro fuzzy technique is utilized to solve the inverse problem of crack identification. Two different neuro fuzzy systems including grid partitioning (GP) and subtractive clustering (SC) are investigated for the highlighted problem. Several error metrics are utilized for evaluating the accuracy of the hybrid algorithms. The study is the first in terms of 1) using the two models of neuro fuzzy systems in crack detection and 2) considering multiple damages in beam elements employing the fused neuro fuzzy procedures. At the end of the study, the developed hybrid models are tested by utilizing the noise-contaminated data. Considering the robustness of the models, they can be employed as damage identification algorithms in health monitoring of beam-like structures.