• Journal of the Korea Computer Graphics Society
• /
• v.12 no.2
• /
• pp.39-44
• /
• 2006

컴퓨터 그래픽스 분야에서 변형체 애니메이션은 매우 중요한 기법으로 그 응용 범위가 매우 넓다. 본 논문에서는 큰 회전 변형에도 부피의 왜곡이 발생하지 않으며, 변형체의 어느 한 부위가 고정되어 있지 않고 자유롭게 움직일 수 있는 동적 탄성 변형을 실시간에 시뮬레이션 하는 기법을 제안한다. Choi와 Ko [3]가 제안한 모달 와핑 기법은 실시간에 부피의 왜곡이 없는 동적 탄성 변형을 생성할 수 있지만 탄성체의 어느 한 부위가 고정 되어 있어야 하며, Hauser 등 [9]이 제안한 임펄스 기반의 충돌 반응을 고려한 선형 모달 해석법은 안정적인 시뮬레이션 결과를 주지만 큰 변형에 있어서 부피 왜곡이 있다. 본 논문에서는 Choi와 Ko [3]가 제안한 모달 와핑 기법을 임펄스 기반의 충돌 반응을 포함하도록 확장하여 큰 회전 변형에도 부피의 왜곡을 방지하며 또한 자유롭게 움직이는 탄성체의 충돌 반응을 안정적으로 시뮬레이션 할 수 있게 한다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2021.07a
• /
• pp.667-669
• /
• 2021

본 논문에서는 벡터 기반 데이터 증강 기법(Data augmentation)을 제안하여 학습 데이터를 구축한 뒤, 이를 합성곱 신경망(Convolutional Neural Networks, CNN)으로 실제 균열과 가까운 패턴을 표현할 수 있는 프레임워크를 제안한다. 건축물의 균열은 인명 피해를 가져오는 건물 붕괴와 낙하 사고를 비롯한 큰 사고의 원인이다. 이를 인공지능으로 해결하기 위해서는 대량의 데이터 확보가 필수적이다. 하지만, 실제 균열 이미지는 복잡한 패턴을 가지고 있을 뿐만 아니라, 위험한 상황에 노출되기 때문에 대량의 데이터를 확보하기 어렵다. 이러한 데이터베이스 구축의 문제점은 인위적으로 특정 부분에 변형을 주어 데이터양을 늘리는 탄성왜곡(Elastic distortion) 기법으로 해결할 수 있지만, 본 논문에서는 이보다 향상된 균열 패턴 결과를 CNN을 활용하여 보여준다. 탄성왜곡 기법보다 CNN을 이용했을 때, 실제 균열 패턴과 유사하게 추출된 결과를 얻을 수 있었고, 일반적으로 사용되는 픽셀 기반 데이터가 아닌 벡터 기반으로 데이터 증강을 설계함으로써 균열의 변화량 측면에서 우수함을 보였다. 본 논문에서는 적은 개수의 균열 데이터를 입력으로 사용했음에도 불구하고 균열의 방향 및 패턴을 다양하게 생성하여 쉽게 균열 데이터베이스를 구축할 수 있었다. 이는 장기적으로 구조물의 안정성 평가에 이바지하여 안전사고에 대한 불안감에서 벗어나 더욱 안전하고 쾌적한 주거 환경을 조성할 것으로 기대된다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.01a
• /
• pp.377-380
• /
• 2023

본 논문에서는 합성곱 신경망(Convolutional Neural Networks, CNN)과 탄성왜곡(Elastic Distortion) 기법을 통한 데이터 증강 기법을 활용하여 학습 데이터를 구축하는 프레임워크를 제안한다. 실제 균열 이미지는 정형화된 형태가 없고 복잡한 패턴을 지니고 있어 구하기 어려울 뿐만 아니라, 데이터를 확보할 때 위험한 상황에 노출될 우려가 있다. 이러한 데이터베이스 구축 문제점을 본 논문에서 제안하는 데이터 증강 기법을 통해 비용적, 시간적 측면에서 효율적으로 해결한다. 세부적으로는 DeepCrack의 데이터를 10배 이상 증가하여 실제 균열의 특징을 반영한 메타 데이터를 생성하여 U-net을 학습하였다. 성능을 검증하기 위해 균열 탐지 연구를 진행한 결과, IoU 정확도가 향상되었음을 확인하였다. 데이터를 증강하지 않았을 경우 잘못 예측(FP)된 경우의 비율이 약 25%였으나, 데이터 증강을 통해 3%까지 감소하였음을 확인하였다.

• Journal of the Korea Computer Graphics Society
• /
• v.28 no.2
• /
• pp.1-9
• /
• 2022

In this paper, we propose a vector-based augmentation technique that can generate data required for crack detection and a ConvNet(Convolutional Neural Network) technique that can learn it. Detecting cracks quickly and accurately is an important technology to prevent building collapse and fall accidents in advance. In order to solve this problem with artificial intelligence, it is essential to obtain a large amount of data, but it is difficult to obtain a large amount of crack data because the situation for obtaining an actual crack image is mostly dangerous. This problem of database construction can be alleviated with elastic distortion, which increases the amount of data by applying deformation to a specific artificial part. In this paper, the improved crack pattern results are modeled using ConvNet. Rather than elastic distortion, our method can obtain results similar to the actual crack pattern. By designing the crack data augmentation based on a vector, rather than the pixel unit used in general data augmentation, excellent results can be obtained in terms of the amount of crack change. As a result, in this paper, even though a small number of crack data were used as input, a crack database can be efficiently constructed by generating various crack directions and patterns.

• Geophysics and Geophysical Exploration
• /
• v.20 no.3
• /
• pp.163-175
• /
• 2017

In land seismic exploration, irregular surface topography and weathering layer in near surface distorts the reflected signals of data. Therefore, typical land seismic data should be compensated for this distortion by static correction. To perform the static correction, near-surface velocity is required, which can be obtained by seismic refraction survey. However, land seismic data is often acquired in a limited form of geometry depending on the equipment availability, accessibility condition, and permission for the survey site. In this situation, refraction analysis should be performed using reflection data because it is impossible to acquire refraction-oriented data due to limited source and receiver geometry. In this study, we aimed to analyze the reliability of the results obtained by refraction traveltime tomography when using reflection data with a limited number of sources and receivers from irregular surface topography. By comparing the inversion result from irregular topography with that from flat surface, we found that the surface topography affects the reliability of the inversion results to some degree. We also found that the number of sources has little effect on the inversion results unless the number of sources are very small. On the other hand, we observed that velocity distortion occurred in the overlapped part of receiver arrays when using a limited number of receivers, and therefore suggested the size of the least overlapping ratio to avoid the velocity distortion. Finally, we performed numerical tests for the model which simulates the surface topography and acquisition geometry of the survey region and verified the reliability analysis of inversion results. We identified reliable areas and suspicious area of the inverted velocity model by applying the analysis results to field data.

• Geophysics and Geophysical Exploration
• /
• v.1 no.1
• /
• pp.31-42
• /
• 1998

Recently crosshole seismic tomography has come to be widely used especially for the civil engineering, because it can provide more detail information than any other surface method, although the resolution of tomogram will be inevitably deteriorated to some extent due to the limited wavefield aperture on the nonuniqueness of traveltime inversion. In addition, our field sites often consist of a high-velocity bed rock overlain by low-velocity rock, sometimes with a contrast of more than 45 percent, and furthermore the bed rock is folded. The first arriving waves can be then the refracted ones that travel along the bed rock surface for some source/receiver distances. Thus, the desirable first arrivals can be easily misread that cause severe distortion of the resulting tomogram, if it is concerned with (straight ray) traveltime inversion procedure. In this case, comparision with synthetic data (forward modeling) is a valuable tool in the interpretation process. Besides, abundant information is contained in the crosshole data. For instance, examination of tube waves can be devoted to detecting discontinuities within the borehole such as breakouts, faults, fractures or shear zones as well as the end of the borehole. Specific frequency characteristics of marine silty mud will help discriminate from other soft rocks. The aim of this paper is to present several strategies to analyze and interpret the crosshole data in order to improve the ability at first to determine the spatial dimensions of interwell anomalies and furthermore to understand the underground structures. To this end, our field data are demonstrated. Possibility of misreading the first arrivals was illustrated. Tube waves were investigated in conjunction with the televiewer images. Use of shot- and receiver gathers was examined to benefit the detectabilities of discontinuities within the borehole.

• Geophysics and Geophysical Exploration
• /
• v.7 no.4
• /
• pp.245-250
• /
• 2004

Horizontal-axes on the seismic section have been represented in a distance unit by applying horizontal-distance correction transformation on a 2-D seismic section of single channel marine seismic data. By drawing horizontal-axes in a distance unit, distortion of horizontal distances shown on the seismic section when the ship speed varies during a survey can be diminished considerably. Position information obtained by GPS and stored in each trace of seismic data as well as data collection windows were used for horizontal distance correction. The minimum window length was decided by considering ship speed and shot interval, and the maximum window length wat determined by reflecting radius of the 1st Fresnel zone. In choosing an optimum window length, horizontal resolution and stacking effect were considered simultaneously. By applying horizontal distance correction we could get a 2-D seismic section which is considered at reflecting the real subsurface structure analogously.

• Geophysics and Geophysical Exploration
• /
• v.6 no.1
• /
• pp.28-34
• /
• 2003

A problem of ground subsidence has been a focus of our research over the past 3 years. The purpose of this study is to investigate the disturbed stratigraphic structure by mining and to separate the possible ground subsidence area using shallow seismic reflection survey and processing. To overcome the problems such as the distortion and attenuation of seismic signal caused by ground disturbance and to acquire the high frequency data, an array with short spacing (0.3m) for both the shot and receivers, yielding near-offset (<30m) and CMP spacing of 0.15m was implemented. Data were acquired along the survey line with length of about 43m by fixed receiver array. By considering statics caused by the ground disturbance and offset distribution of data, careful processing steps such as muting and residual statics correction were applied for successful shallow reflection imaging. By correlating the ground subsidence data and stack section, possible subsidence zone could be interpreted quantitatively.

• Geophysics and Geophysical Exploration
• /
• v.5 no.3
• /
• pp.145-149
• /
• 2002

The static correction, which is classified into refraction based static correction and reflection based residual static correction, removes distortions caused by irregularities of thickness or velocity in near-surface. Generally, refraction statics is a time consuming process because of high dependence on the interpreter's analysis. Therefore, for huge 3D seismic data, automatic static correction which minimizes the interpreter's analysis is required. In this research, we introduce an efficient method of refraction static correction for land 3D seismic survey.

• Journal of the Korean Geophysical Society
• /
• v.6 no.1
• /
• pp.31-38
• /
• 2003

Multi-channel seismic survey, which has been mainly employed in oil prospecting, is carried out as a high resolution shallow marine seismic exploration. Fault drop as small as 1 m can be resolved by employing high-resolution seismic survey. Similar to the effect of shallow inhomogenities in the land seismic data, due to occurrence of swell quite often higher than 1 m, shallow marine seismic data tend to be severely degraded. Suppression of such a swell effect is critical in processing of steps of marine seismic shallow high-resolution data. Compared to the moving average depth method, a newly developed method using cross-correlation technique is found out to be very effective in increasing the resolution of the shallow reflection events by accuratly elucidating the depth of sea bottom.