• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.8
• /
• pp.2787-2800
• /
• 2022

The aging society increases emergency situations of the elderly living alone and a variety of social crimes. In order to prevent them, techniques to detect emergency situations through voice are actively researched. This study proposes CutPaste-based anomaly detection model using multi-scale feature extraction in time series streaming data. In the proposed method, an audio file is converted into a spectrogram. In this way, it is possible to use an algorithm for image data, such as CNN. After that, mutli-scale feature extraction is applied. Three images drawn from Adaptive Pooling layer that has different-sized kernels are merged. In consideration of various types of anomaly, including point anomaly, contextual anomaly, and collective anomaly, the limitations of a conventional anomaly model are improved. Finally, CutPaste-based anomaly detection is conducted. Since the model is trained through self-supervised learning, it is possible to detect a diversity of emergency situations as anomaly without labeling. Therefore, the proposed model overcomes the limitations of a conventional model that classifies only labelled emergency situations. Also, the proposed model is evaluated to have better performance than a conventional anomaly detection model.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.6
• /
• pp.303-310
• /
• 2023

A single vision system limits the ability to accurately understand the spatial constraints and interactions between robots and dynamic workers caused by gantry robots and collaborative robots during production manufacturing. In this paper, we propose a 3D pose registration method for dynamic workers based on a multi-domain vision system for safety monitoring in manufacturing environments. This method uses OpenPose, a deep learning-based posture estimation model, to estimate the worker's dynamic two-dimensional posture in real-time and reconstruct it into three-dimensional coordinates. The 3D coordinates of the reconstructed multi-domain vision system were aligned using the ICP algorithm and then registered to a single 3D coordinate system. The proposed method showed effective performance in a manufacturing process environment with an average registration error of 0.0664 m and an average frame rate of 14.597 per second.

• Journal of Oral Medicine and Pain
• /
• v.41 no.3
• /
• pp.85-90
• /
• 2016

Purpose: The aim of this study was to develop an automated software to extract tooth and pulpal area from sectional cone-beam computed tomography (CBCT) images, which can guarantee more reproducible, objective and time-saving way to measure pulp/tooth volume ratio. Methods: The software program was developed using MATLAB (MathWorks). To determine the optimal threshold for the region of interest (ROI) extraction, user interface to adjust the threshold for extraction algorithm was added. Default threshold was determined after several trials to make the outline of extracted ROI fitting to the tooth and pulpal outlines. To test the effect of starting point location selected initially in the pulpal area on the final result, pulp/tooth volume ratio was calculated 5 times with different 5 starting points. Results: Navigation interface is composed of image loading, zoom-in, zoom-out, and move tool. ROI extraction process can be shown by check in the option box. Default threshold is adjusted for the extracted tooth area to cover whole tooth including dentin, cementum, and enamel. Of course, the result can be corrected, if necessary, by the examiner as well as by changing the threshold of density of hard tissue. Extracted tooth and pulp area are reconstructed three-dimensional (3D) and pulp/tooth volume ratio is calculated by voxel counting on reconstructed model. The difference between the pulp/tooth volume ratio results from the 5 different extraction starting points was not significant. Conclusions: In further studies based on a large-scale sample, the most proper threshold to present the most significant relationship between age and pulp/tooth volume ratio and the tooth correlated with age the most will be explored. If the software can be improved to use whole CBCT data set rather than just sectional images and to detect pulp canal in the original 3D images generated by CBCT software itself, it will be more promising in practical uses.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.11 no.1
• /
• pp.7-18
• /
• 1993

In analyzing whole surface of non-topographic objects, the design of multi-station photogrammetric network must involve a number of questions such as geometric configuration of exposure stations, imaging geometry, control point configuration or weight allowance of adjustments. Above all, the surveying of the interior of narrow longitudinal structures needs the design of special photogrammetric network. The main objective of this paper is to suggest the schemes for solving difficult problems attendant upon whole inside-surface analysis of structure and to improve the accuracy and reliability of final measurements. For it, the multi-station exposure network suitable to shape and size of the inside of the structure was designed. Then three dimensional data were acquired by bundle adjustments derived from multi-station photos and the effects of network design factors on accuracy of measurements were contemplated. Also, the algorithm for detection of blunders was developed here is expected to lead to improvement of the reliability of photogrammetric solutions.

• The KIPS Transactions:PartB
• /
• v.16B no.5
• /
• pp.359-366
• /
• 2009

Analysis of 3-dimensional (3D) protein structure plays an important role of structural bioinformatics. The protein structure alignment is the main subjects of the structural bioinformatics and the most fundamental problem. Protein Structures are flexible and undergo structural changes as part of their function, and most existing protein structure comparison methods treat them as rigid bodies, which may lead to incorrect alignment. We present a new method that carries out the flexible structure alignment by means of finding SSPs(Similar Substructure Pairs) and flexible points of the protein. In order to find SSPs, we encode the coordinates of atoms in the backbone of protein into RDA(Relative Direction Angle) using local similarity of protein structure. We connect the SSPs with Floyd-Warshall algorithm and make compatible SSPs. We compare the two compatible SSPs and find optimal flexible point in the protein. On our well defined performance experiment, 68 benchmark data set is used and our method is better than three widely used methods (DALI, CE, FATCAT) in terms of alignment accuracy.

• Proceedings of the Korean Society of Medical Physics Conference
• /
• 2002.09a
• /
• pp.150-153
• /
• 2002

High dose rate (HDR) brachytherapy in the treatment of cervix carcinoma has become popular, because it eliminated many of the problems with conventional brachytherapy. In order to improve clinical effectiveness with HDR brachytherapy, dose calculation algorithm, optimization procedures, and image registrations should be verified by comparing the dose distributions from a planning computer and those from a humanoid phantom irradiated. Therefore, the humanoid phantom should be designed such that the dose distributions could be quantitatively evaluated by utilizing the dosimeters with high spatial resolution. Therefore, the small size of thermoluminescent dosimeter (TLD) chips with the dimension of 1/8" and film dosimetry with spatial resolution of <1mm used to measure the radiation dosages in the phantom. The humanoid phantom called a pelvic phantom is made of water and tissue-equivalent acrylic plates. In order to firmly hold the HDR applicators in the water phantom, the applicators are inserted into the grooves of the applicator supporters. The dose distributions around the applicators, such as Point A and B, can be measured by placing a series of TLD chips (TLD-to- TLD distance: 5mm) in three TLD holders, and placing three verification films in orthogonal planes.

• Journal of Biomedical Engineering Research
• /
• v.30 no.6
• /
• pp.461-468
• /
• 2009

Magnetic resonance electrical impedance tomography (MREIT) enables us to perform high-resolution conductivity imaging of an electrically conducting object. Injecting low-frequency current through a pair of surface electrodes, we measure an induced magnetic flux density using an MRI scanner and this requires a sophisticated MR phase imaging method. Applying a conductivity image reconstruction algorithm to measured magnetic flux density data subject to multiple injection currents, we can produce multi-slice cross-sectional conductivity images. When there exists a local region of fat, the well-known chemical shift phenomenon produces misalignments of pixels in MR images. This may result in artifacts in magnetic flux density image and consequently in conductivity image. In this paper, we investigate chemical shift artifact correction in MREIT based on the well-known three-point Dixon technique. The major difference is in the fact that we must focus on the phase image in MREIT. Using three Dixon data sets, we explain how to calculate a magnetic flux density image without chemical shift artifact. We test the correction method through imaging experiments of a cheese phantom and postmortem canine head. Experimental results clearly show that the method effectively eliminates artifacts related with the chemical shift phenomenon in a reconstructed conductivity image.

• The Journal of the Acoustical Society of Korea
• /
• v.16 no.2
• /
• pp.41-48
• /
• 1997

This paper proposes the methods for improving the recognition rate of theARS, especially equipped with the speech recognition capability. Telephone speech, which is the input to the ARS, is usually affected by the announcements from the system, channel noise, and channel distortion, thus directly applying the recognition algorithm developed for clean speech to the noisy telephone speech will bring the significant performance degradation. To cope with this problem, this paper proposes three methods: 1)the accurate detection of the inputting instant of the speech in order to immediately turn off the announcements from the system at that instant, 2)the effective end-point detection of the noisy telephone speech on the basis of Teager energy, and 3)the SDCN-based compensation of the channel distortion. Experiments on speaker-independent, noisy telephone speech reveal that the combination of the above three proposed methods provides great improvements on the recognition rate over the conventional method, showing about 77% in contrast to only 23%.

• Tunnel and Underground Space
• /
• v.13 no.6
• /
• pp.486-494
• /
• 2003

When the orientations of joints are measured on a rock exposure, there are frequent cases that are difficult to approach by the surveyor to the target joints or to set up scanlines on the slope. In this study, to complement such limit and weak points, a new algorithm was developed to interpret joint orientation from analyzing the images of rock slope. As a method of arranging the multiple images of a rock slope, the multistage convergent photographing system was introduced to overcome the limitation of photographing direction which existing method such as parallel stereophotogrammetric system has and to cover the range of image measurement, which is the overlapping area between the image pair, to a maximum extent. To determine camera parameters in the perspective projection equation that are the main elements of the analysis method, a new method was developed introducing three ground control points and single ground guide point. This method could be considered to be very simple compared with other existing methods using a number of ground control points and complicated analysis process. So the global coordinates of a specific point on a rock slope could be analyzed with this new method. The orientation of a joint could be calculated using the normal vector of the joint surface which can be derived from the global coordinates of several points on the joint surface analyzed from the images.

• Korean Journal of Remote Sensing
• /
• v.14 no.3
• /
• pp.250-261
• /
• 1998

The purpose of this paper is to extract fast DEM (Digital Elevation Model) using satellite images. DEM extraction consists of three parts. First part is the modeling of satellite position and attitude, second part is the matching of two images to find corresponding points of them and third part is to calculate the elevation of each point by using the results of the first and second part. The position and attitude modeling of satellite is processed by using GCPs. A area based matching method is used to find corresponding points between the stereo satellite images. The elevation of each point is calculated using the exterior orientation parameters obtained from modeling and conjugate points from matching. In the DEM generation system, matching procedure holds most of a processing time, therefore to reduce the time for matching, a new fast matching algorithm using gradient correlation and fast similarity measure calculation method is proposed. In this paper, the SPOT satellite images, level 1A 6000$\times$6000 panchromatic images are used to extract DEM. The experiment result shows the possibility of fast DEM extraction with the satellite images.