• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.29B no.7
• /
• pp.70-78
• /
• 1992

In this paper, an approach for 3-D object segmentation and classification, which is based on eigen-values of polynomial function as their surface features, using neural network is proposed. The range images of 3-D objects are classified into surface primitives which are homogeneous in their intrinsic eigenvalue properties. The misclassified regions due to noise effect are merged into correct regions satisfying homogeneous constraints of Hopfield neural network. The proposed method has advantage of processing both segmentation and classification simultaneously.

• KIEE International Transaction on Electrical Machinery and Energy Conversion Systems
• /
• v.3B no.3
• /
• pp.122-127
• /
• 2003

This paper presents a new time-stepping 3-D analysis method coupled with an external circuit with motion equation for dynamic transient analysis of induction machines. In this method, the magneto-motive force (MMF) generated by induced current is modeled as a passive source in the magnetic equivalent network. So, by using only scalar potential at each node, the method is able to analyze induction machines with faster computation time and less memory requirement than conventional numerical methods. Also, this method is capable of modeling the movement of the mover without the need for re-meshing and analyzing the time harmonics for dynamic characteristics. From comparisons between the results of the analysis and the experiments, it is verified that the proposed method is capable of estimating the torque, harmonic field, etc. as a function of time with superior accuracy.

• Journal of Sensor Science and Technology
• /
• v.28 no.3
• /
• pp.182-186
• /
• 2019

When a camera is employed for 3D sensing, accurate camera calibration is vital as it is a prerequisite for the subsequent steps of the sensing process. Camera calibration is usually performed by complex mathematical modeling and geometric analysis. On the other contrary, data learning using an artificial neural network can establish a transformation relation between the 3D space and the 2D camera image without explicit camera modeling. However, a neural network requires a large amount of accurate data for its learning. A significantly large amount of time and work using a precise system setup is needed to collect extensive data accurately in practice. In this study, we propose a two-step neural calibration method that is effective when only a small amount of learning data is available. In the first step, the camera projection transformation matrix is determined using the limited available data. In the second step, the transformation matrix is used for generating a large amount of synthetic data, and the neural network is trained using the generated data. Results of simulation study have shown that the proposed method as valid and effective.

• 한국정보통신설비학회:학술대회논문집
• /
• 2004.08a
• /
• pp.328-333
• /
• 2004

SK Telecom은 2000년 10월 세계 최초로 기존 IS-95A/B망에서 지원하였던 속도인 14.4Kbps나 56Kbps 보다 훨씬 빠른 최고 144Kbps로 무선인터넷이 가능한 CDMA2000 IX를 상용화 하였다. 또한, 2002년 1월에는 CDMA2000 IX보다 15배 이상 빠른 최대 2.4Mbps가 가능한 동기식 3세대 망인 CDMA2000 IxEV-DO를 세계 최초로 상용화 하였다. 이러한 초고속 무선 인터넷 서비스를 위해서 패킷 데이터 처리에 필수적인 PDSN(Packet Data Serving Node), HA(Home Agent), AAA(Authentication, Authorization and Accounting) 등의 노드들로 구성된 Packet Core Network(이하 PCN)이 도입되었으며, 이에 대한 운용 및 관리 방안이 중요한 issue로 등장하였다. 본 논문에서는 기존의 음성 서비스 관리를 위한 망관리 시스템(Network Management System)과는 다른 개념으로 관리되어야 할 패킷 서비스를 위한 NMS 구축 방안을 제시하고, 필수적인 관리 정보 및 서비스 관리를 위한 방향을 제시한다.

• Journal of Broadcast Engineering
• /
• v.23 no.5
• /
• pp.614-621
• /
• 2018

This paper proposes a method of restoring corrupted depth image captured by depth camera through unsupervised learning using generative adversarial network (GAN). The proposed method generates restored face depth images using 3D morphable model convolutional neural network (3DMM CNN) with large-scale CelebFaces Attribute (CelebA) and FaceWarehouse dataset for training deep convolutional generative adversarial network (DCGAN). The generator and discriminator equip with Wasserstein distance for loss function by utilizing minimax game. Then the DCGAN restore the loss of captured facial depth images by performing another learning procedure using trained generator and new loss function.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.3
• /
• pp.228-233
• /
• 2021

In this paper, we propose a method to extract and classify features of multi-channel ElectroEncephalo Graphy (EEG) that change according to various musical tempo stimuli. In the proposed method, a 3D convolutional bidirectional gated recurrent neural network extracts spatio-temporal and long time-dependent features from the 3D EEG input representation transformed through the preprocessing. The experimental results show that the proposed tempo stimuli classification method is superior to the existing method and the possibility of constructing a music-based brain-computer interface.

• International Journal of Computer Science & Network Security
• /
• v.21 no.1
• /
• pp.152-161
• /
• 2021

The paper focuses on Device-to-device (D2D) Architectures evaluation frameworks. D2D communication and discovery can improve spectrum usage efficiency and optimize the tradeoffs between throughput and energy consumption. The target operation modes involve both indirect communication between two nodes via a base station or the direct communication among proximal nodes, enabling use cases that can support communications out of cellular coverage, as well as low end-end delay requirements. The paper will present the architectural evolution of D2D networks within 3GPP standardization and will highlight key network functionalities and signaling protocols. It will also identify key analytical and simulation models that can be used to assess the performance and energy efficiency of resource allocation strategies, and it will present a suitable cross-layer integrated framework.

• International Journal of Internet, Broadcasting and Communication
• /
• v.11 no.2
• /
• pp.31-42
• /
• 2019

As smartphone H/W performance and mobile communication service have been enhanced, large-capacity 3D modeling files are available in smartphones. Common formats of 3D modeling files include STL (STereoLithography), OBJ (Wavefront file format specification), FBX (Filmbox), and glTF (open GL Transmission Format). Each format has different characteristics depending on the configuration and functions, and formats that are supported are varied depending on the applications. Large-size files are commonly used. The 4th generation mobile communication network secures loading of 3D modeling files and transmission of large-size geometric files in order to provide augmented reality services via smartphones. This paper explains the concepts and characteristics of major 3D file formats such as OBJ, FBX, and glTF. In addition, it compares their performance in a wired web with that in the 4th generation mobile communication network. The loading time and packet transmission in each 3D format are also measured by means of different mobile web browsers (Google Chrome and MS Edge). The experiment result shows that glTF demonstrated the most efficient performance while the loading time of OBJ was relatively excessive. Findings of this study can be utilized in selecting specific 3D file formats for rendering time reduction depending on the mobile web environments.

• The Journal of the Korea Contents Association
• /
• v.12 no.9
• /
• pp.48-56
• /
• 2012

The interaction using digital media has been widely utilized in modern art. In particular, network based digital media enables two-way and real-time communication regardless of time and place. Network is possible to connect to effective and meaningful communication of extended relationship. This study proposes novel human-MediaFacade interaction method through a mechanism on interworking between 3D mapping technology and social media, smart devices. This study has an important meaning which can be set multi-user interaction. And a study of creating 'Facade' which interaction has high artistic value.

• Journal of Information Processing Systems
• /
• v.15 no.5
• /
• pp.1179-1191
• /
• 2019

We design an ingenious view-pooling method named learning-based multiple pooling fusion (LMPF), and apply it to multi-view convolutional neural network (MVCNN) for 3D model classification or retrieval. By this means, multi-view feature maps projected from a 3D model can be compiled as a simple and effective feature descriptor. The LMPF method fuses the max pooling method and the mean pooling method by learning a set of optimal weights. Compared with the hand-crafted approaches such as max pooling and mean pooling, the LMPF method can decrease the information loss effectively because of its "learning" ability. Experiments on ModelNet40 dataset and McGill dataset are presented and the results verify that LMPF can outperform those previous methods to a great extent.