• Electronics and Telecommunications Trends
• /
• v.33 no.4
• /
• pp.1-13
• /
• 2018

Deep learning (DL) models have been widely applied to AI applications such image recognition and language translation with big data. Recently, DL models have becomes larger and more complicated, and have merged together. For the accelerated training of a large-scale deep learning model, model parallelism that partitions the model parameters for non-shared parallel access and updates across multiple machines was provided by a few distributed deep learning frameworks. Model parallelism as a training acceleration method, however, is not as commonly used as data parallelism owing to the difficulty of efficient model parallelism. This paper provides a comprehensive survey of the state of the art in model parallelism by comparing the implementation technologies in several deep learning frameworks that support model parallelism, and suggests a future research directions for improving model parallelism technology.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.6
• /
• pp.247-252
• /
• 2004

This paper proposes a matching algorithm using Voronoi diagram for rotation and translation invariant fingerprint identification. The proposed algorithm extracts geometrical structures that are derived from Voronoi diagram of a fingerprint image. Then two features, distances and angles are extracted from the geometrical structures and saved as indexing form for fingerprint matching. This matching algerian has a lower error rate than indexing based methods of old times.

• 한국전산유체공학회:학술대회논문집
• /
• 2007.10a
• /
• pp.67-72
• /
• 2007

A post-processing program based on the OOP(Object-Oriented Programming) concept has been developed for visualization of MRI. User-friendly GUl(Graphic User Interface) has been built on the base of MFC(Microsoft Foundation Class). The program is organized as modules by classes based on VTK-library, and these classes are made to function through inheritance and cooperation which are an important and valuable concept of object-oriented programming. The major functions of this post-processor program are introduced and demonstrated, which include contour plot, surface plots, cut plot and clip plot as well as view manipulation (translation, rotation, scaling etc).

• Proceedings of the KSME Conference
• /
• 2001.11a
• /
• pp.192-197
• /
• 2001

To obtain $A_2$ experimentally in the $J-A_2$ theory, deformation field on the lateral surface of a CT specimen was to be determined using Laser speckle method. The crack growth was measured using direct current potential drop method and most procedure of experimental and data reduction was performed according to ASTM Standard E1737-96. Laser speckle images during crack propagation were monitored by two CCD cameras to cancel the effect of rotation and translation of the specimen. An algorithm to pursue displacement of a point from each image was developed and successfully used to measure $A_2$ continuously as the crack tip was propagated. The effects of specimen thickness on J-R curve and $A_2$ were explored.

• Journal of the Korean Society of Manufacturing Process Engineers
• /
• v.13 no.2
• /
• pp.28-34
• /
• 2014

A non-contact type measurement system is specially devised to measure the width of a piston ring groove in the grinding process. This system comprises a line camera with an imaging lens, collimated white light source, and a one axis translation stage. When the measurement system movesalong the diagonal direction of the cylinder, the line camera captures an image. By analyzing such images, the width of the piston ring groove can be determined. The experimental results prove that the proposed system is useful, especially as a monitoring system in grinding piston ring grooves on cylinders with accuracy of several micrometers in an area of dozens of millimeters.

• Journal of computational fluids engineering
• /
• v.19 no.2
• /
• pp.30-39
• /
• 2014

We report in this paper the analysis of the motion of a needle and a spring in a viscous fluid under the influence of gravitational force. Lateral shift as well as vertical motion of a needle falling in a viscous fluid has been observed from a simple experiment. We also observed the combined rotation and translation of a falling spring. The trajectory and velocity of the falling needle and the spring were obtained by using an image processing technique. We also conducted numerical simulation for both problems. For the falling-needle problem, we employed a theory; but it turns out that significant correction is required for the solutions to match the numerical and experimental data. For the falling spring problem various theoretical formula were tested for their justification, but none of the existing theories can successfully predict the numerical and experimental results.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2016.04a
• /
• pp.166-167
• /
• 2016

CCTV는 다양한 목적을 위해 24시간 영상을 녹화하지만 현재 시스템으로 사각지대 없는 CCTV 시스템은 다수의 카메라를 설치하여만 한다. 최근 360도 CCTV의 등장으로 동시에 사각지대 없는 촬영이 가능하지만 아직은 해상도, 이미지 변환과 같은 기술적인 개선 연구가 지속적으로 보완되어야 하는 상황이다. 본 논문에서는 천정, 바닥, 벽에 부착된 360도 CCTV 촬영된 이미지를 안드로이드 모바일 기기에서 변환할 수 있는 앱을 개발하여 이동중에 촬영영상을 모니터링할 수 있도록 한다.

• Proceedings of the KIEE Conference
• /
• 2000.07d
• /
• pp.3053-3055
• /
• 2000

The Cellular Nonlinear Networks structure for Distance Transform (DT) and the robust optical flow detection algorithm based on the DT are proposed. The proposed algorithm is for detecting the optical flows on the trajectories only of the feature points. The translation lengths and the directions of feature movements are detected on the trajectories of feature points on which Distance Transform Field is developed. The robustness caused from the use of the Distance Transform and the easiness of hardware implementation with local analog circuits are the properties of the proposed structure, To verify the performance of the proposed structure and the algorithm, simulation has been done about zooming image.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.9
• /
• pp.1171-1180
• /
• 1995

In this paper, we present a 3-D object recognition system in which the 3-D Hough transform domain is employed to represent the 3-D objects. In object modeling step, the features for recognition are extracted from the CAD models of objects to be recognized. Since the approach is based on the CAD models, the accuracy and flexibility are greatly improved. In matching stage, the sensed image is compared with the stored model, which is assumed to yield a distortion (location and orientation) in the 3-D Hough transform domain. The high dimensional (6-D) parameter space, which defines the distortion, is decomposed into the low dimensional space for an efficient recognition. At first we decompose the distortion parameter into the rotation parameter and the translation parameter, and the rotation parameter is further decomposed into the viewing direction and the rotational angle. Since we use the 3-D Hough transform domain of the input images directly, the sensitivity to the noise and the high computational complexity could be significantly alleviated. The results show that the proposed 3-D object recognition system provides a satisfactory performance on the real range images.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2001.10a
• /
• pp.739-742
• /
• 2001

영상처리 기법을 이용하여 움직이는 물체를 추적하는 시스템은 많은 분야에 응용될 수 있다. 기존의 물체 추적 시스템에서 사용한 방법들은 최적의 특성을 지니고 있지만, 많은 계산량을 요구하는 단점이 있다. 또한 단안에 의한 영상의 경우 물체의 다양한 변화에 대한 예측과 추적이 어렵다. 따라서 본 논문에서는 양안에 의해 얻어진 스테레오 영상에 전이(translation)와 회전(rotation)에서의 예측이 어려운 단점을 보완한 블록 정합 알고리즘을 적용함으로써 실시간 물체의 변화의 추적능력을 지니면서도 그 계산량을 줄일 수 있는 추적 방법에 대하여 기술한다. 또한 히스토그램의 계수 변화를 이용함으로써 블록 정합 알고리즘에서의 탐색영역을 제한하도록 하여 불필요한 처리시간을 절약하는 효율적인 물체추적 시스템을 제안한다.