• 로봇학회논문지
• /
• 제12권2호
• /
• pp.184-193
• /
• 2017

This work shows how to create an algorithm and implementation for motion and image matching between a vehicle simulator and Unity 3D based virtual object. The motion information of the virtual vehicle is transmitted to the real simulator via a RS232 communication protocol, and the motion is controlled based on the inverse kinematics solution of the platform adopting rotary-type six actuators driving system. Wash-out filters to implement the effective motion of the motion platform are adopted, and thereby reduce the dizziness and increase the realistic sense of motion. Furthermore, the simulator system is successfully designed aiming to reducing size and cost with adaptation of rotary-type six actuators, real driving environment via VR (Virtual Reality), and control schemes which employ a synchronization between 6 motors and 3rd order motion profiles. By providing relatively big sense of motion particularly in impact and straight motions mainly causing simulator sickness, dizziness is remarkably reduced, thereby enhancing the sense of realistic motion.

• 방송공학회논문지
• /
• 제13권6호
• /
• pp.915-929
• /
• 2008

본 논문은 다시점 카메라로부터 획득된 영상을 이용하여 영상내의 모든 화소에 대한 정확한 변이 정보를 구하는 알고리듬을 제안한다. 제안한 방법은 객체의 경계 정보를 고려하여 초기 변이를 예측한 후 획득된 변이 정보를 이용하여 탐색 범위를 줄임으로 써 효율적으로 변이를 예측한다. 또한 가변 블록을 사용하여 텍스쳐 정보가 부족한 영역과 경계부분에서 발생하는 오정합 문제를 줄일 수 있다. 획득된 변이 맵 정보를 이용하여 중간시점영상을 생성한 결과 기존의 블록기반 변이 추정방식과 화소기반의 변이 예측방식에 비해 $0.1dB{\sim}1.2dB$의 PSNR(Peak signal to noise ratio)이 향상되는 것을 확인하였다.

• 정보처리학회논문지B
• /
• 제14B권2호
• /
• pp.99-106
• /
• 2007

본 논문에서는 적응적 관심영역(AAW: Adaptive Attention Window)에 기반한 세포영상 분할 기법을 제안한다. 적응적 관심영역은 분할하기 위해, 명암지도를 이용하여 초기 관심윈도우(IAW: Initial AW)를 생성한다. 생성된 초기 관심윈도우는 쿼드-트리 분할을 이용하여 실제의 관심영역(ROI: Region of Interest)과 유사한 크기가 될 때까지 축소된다. 이렇게 생성된 적응적 관심윈도우는 세포 영상에서 배경을 제거하고 관심영역 추출의 처리 시간을 줄이기 위해서 사용된다. 마지막으로 적응적 관심영역 안에서 영역을 분할하고, 관심영역만을 분리하기 위한 영역 병합과 제거를 수행한다. 실험에서 제안된 기법은 세포영상의 관심영역을 효과적으로 분리하여 인간 시각과 유사한 향상된 영상 분할 결과를 보여준다.

• 한국통신학회논문지
• /
• 제36권12C호
• /
• pp.727-737
• /
• 2011

본 논문에서는 FPGA(Field Programmable Gate Array)와 DSP(Digital Signal Processor)를 이용하는 실시간 차선 및 차량인식 시스템의 구현에 대하여 기술한다. 실시간 시스템의 구현을 위해서 FPGA와 DSP의 역할을 효율적으로 분할할 필요성이 있다. 시스템의 알고리즘을 특정요소 추출부분을 기준으로 분할하여 대량의 영상정보를 이용하여 소량의 특정요소를 추출하는 과정을 FPGA로 구현하고 추출된 특정요소를 사용하여 차선과 차량을 정의하고 추적하는 부분을 DSP에서 수행하게 하고, FPGA와 DSP의 효율적 연동을 위한 인터페이스 구성을 제안함으로써 실시간 처리가 가능한 시스템 구조를 제안한다. 실험 결과 제안한 실시간 차선 및 차량인식 시스템은 $640{\times}480$ 크기를 갖는 비디오 영상 입력에 대해 약 15 (frames/sec)로 동작하여 실시간 응용으로 충분함을 알 수 있다.

• 한국컴퓨터정보학회지
• /
• 제2권1호
• /
• pp.130-139
• /
• 1995

본 논문에서는 복원 영상의 화질을 최대한 유지시키면서 신속한 부호화가 가능하도록 에지 추출법 및 부영상 분할 방법을 도입하여 탐색 영역을 줄이는 방법을 제안하였다. 즉 원영상을 부영상으로 분할하므로써 탐색영역인 Domain영역을 1/64까지 줄였으며 에지 추출법으로 에지인 부분과 아닌 부분으로 나누어 같은 클레스에 있는 영역에서만 탐색하도록 하고 그외의 경우는 탐색 영역에서 제외시킴으로서 계산량을 줄였다. 분할된 부영상중 화질이 저하되는 부분은 부영상에 포함된 에지의 임계치에 따라 검색 방법을 달리함으로써 화질 개선을 시도하였다. 또한 부호화시 Range 블록의 크기를 $4{\times}4$와 $8{\times}8$로 했을 때의 압축율과 화질을 비교하여 보았다.

• 한국가시화정보학회지
• /
• 제17권2호
• /
• pp.48-57
• /
• 2019

A straightforward strategy for particle image velocimetry (PIV) interrogation and post-processing has been proposed, aiming at reducing errors and clarifying vortex structures. The interrogation window size should be kept small to reduce bias error and improve spatial resolution. A spatial filter is then applied to the velocity field to reduce random error and clarify flow structure. The performance of three popular spatial filters were assessed: box filter, median filter, and local quadratic polynomial regression filter. In order to quantify random uncertainty, the image matching (IM) method is applied to an experimental dataset of homogeneous and isotropic turbulence (HIT) obtained by 2D-PIV. We statistically analyze the uncertainty propagation through the spatial filters, and verify the reduction in random uncertainty. Moreover, we illustrate that the spatial filters help clarify vortex structures using vortex identification criteria. As a result, PIV random uncertainty was reduced and the vortex structures became clearer by spatial filtering.

• 소성∙가공
• /
• 제31권4호
• /
• pp.229-239
• /
• 2022

Convolution Neural Network(CNN) is a class of deep learning algorithms and can be used for image analysis. In particular, it has excellent performance in finding the pattern of images. Therefore, CNN is commonly applied for recognizing, learning and classifying images. In this study, the surface defect classification performance of Al 6061 extruded material using CNN-based algorithms were compared and evaluated. First, the data collection criteria were suggested and a total of 2,024 datasets were prepared. And they were randomly classified into 1,417 learning data and 607 evaluation data. After that, the size and quality of the training data set were improved using data augmentation techniques to increase the performance of deep learning. The CNN-based algorithms used in this study were VGGNet-16, VGGNet-19, ResNet-50 and DenseNet-121. The evaluation of the defect classification performance was made by comparing the accuracy, loss, and learning speed using verification data. The DenseNet-121 algorithm showed better performance than other algorithms with an accuracy of 99.13% and a loss value of 0.037. This was due to the structural characteristics of the DenseNet model, and the information loss was reduced by acquiring information from all previous layers for image identification in this algorithm. Based on the above results, the possibility of machine vision application of CNN-based model for the surface defect classification of Al extruded materials was also discussed.

• 한국연소학회지
• /
• 제7권4호
• /
• pp.1-9
• /
• 2002

The morphology of deposits on $15-{\mu}m$ thin SiC filaments has been investigated with SEM and compared with UV-excited laser induced broadband fluorescences in co-flowing, propane laminar diffusion flames in a reduced oxidizer environment. The homogeneous morphology of droplet-like deposits inner flame zone and the agglomeration of condensed-phase deposits and the transition to soots from grown up droplet-like precursors with approaching the flame surface can be observed in a barely sooting flame. The average size of the mature soots deposited in the luminous flame edge is scarcely dependent on their axial position in a confined flame under reduced oxidizer condition. A double structure of PAH fluorescence is observed in near-extinction flames with further decreasing of oxidizer. A comparison of the PAH fluorescence with the morphologies of deposits indicates that appearance of the "dark" hollow zone is caused by a decreased number density of developed liquid-phase large molecules and the outer weak fluorescence zone is caused by the diffusion of gas-phase small molecules.

• 척추신경추나의학회지
• /
• 제5권1호
• /
• pp.101-110
• /
• 2010

Objectives: The propose of this study is to find out the clinical application of conservative treatment to 3 patients who has Disc Extrusion on L-spine MRI Methods: We examined 3 patients with Lumbar Intervertebral Disc Herniation (HIVD of L-spine) with Disc Extrusion who showed changes on MRI images before/after the treatment among HIVD of L-spine patients who visited Jaseng Hospital of Oriental Medicine. Results: In this study, the first MRI examination of HIVD of L-spine patients was performed at the first visit and re-examination of MRI was done after the treatment. In each case, the size of the extruded disc was considerably reduced in MRI image. Low back and leg pain was also reduced significantly after conervative oriental medical treatment. Conclusions: Conservative oriental medical treatment can be effective for improving symptoms of HIVD, decreasing pain, also decreasing the volume of herniated disc.

• Smart Structures and Systems
• /
• 제30권5호
• /
• pp.501-511
• /
• 2022

Three-dimensional (3D) models have become crucial for improving civil infrastructure analysis, and they can be used for various purposes such as damage detection, risk estimation, resolving potential safety issues, alarm detection, and structural health monitoring. 3D point cloud data is used not only to make visual models but also to analyze the states of structures and to monitor them using semantic data. This study proposes automating the generation of high-quality 3D point cloud data and removing noise using deep learning algorithms. In this study, large-format aerial images of civilian infrastructure, such as cut slopes and dams, which were captured by drones, were used to develop a workflow for automatically generating a 3D point cloud model. Through image cropping, downscaling/upscaling, semantic segmentation, generation of segmentation masks, and implementation of region extraction algorithms, the generation of the point cloud was automated. Compared with the method wherein the point cloud model is generated from raw images, our method could effectively improve the quality of the model, remove noise, and reduce the processing time. The results showed that the size of the 3D point cloud model created using the proposed method was significantly reduced; the number of points was reduced by 20-50%, and distant points were recognized as noise. This method can be applied to the automatic generation of high-quality 3D point cloud models of civil infrastructures using aerial imagery.