• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.4 no.2
• /
• pp.505-516
• /
• 2000

Cellular Neural Networks(CNN) is a nonlinear information-processing system that has a locally connected characteristic and is widely used in the real-time high speed image processing. In this paper, a practical system approach of time-multiplexing CNN implementations suitable for processing large and complex images using small CNN arrays is presented and $6\times6$ CNN hardware is designed for the processing of a large image. While previous implementations are mostly suitable for black and white applications because of the thresholded outputs, our approach is especially suitable for applications in gray image processing due to the analog nature of the state node. CNN chip is designed using a 0.65${\mu}{\textrm}{m}$ 2P2M(double poly, double metal) N-Well CMOS process technology. It contains about 15,400 devices on an area of about $1.85\times1.75$ md. The designed $6\times6$ CNN is tested for the edge detection of a large image input and it's performance is verified.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.19 no.6
• /
• pp.227-233
• /
• 2019

Recently, along with the recent development of deep learning technique, neural networks are achieving success in computer vision filed. Convolutional neural network have shown outstanding performance in not only for a simple image classification task, but also for tasks with high difficulty such as object segmentation and detection. However many such deep learning models are based on supervised-learning, which requires more annotation labels than image-level label. Especially image semantic segmentation model requires pixel-level annotations for training, which is very. To solve these problems, this paper proposes a weakly-supervised semantic segmentation method which requires only image level label to train network. Existing weakly-supervised learning methods have limitations in detecting only specific area of object. In this paper, on the other hand, we use multi-classifier deep learning architecture so that our model recognizes more different parts of objects. The proposed method is evaluated using VOC 2012 validation dataset.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.8 no.1
• /
• pp.24-29
• /
• 2003

A 120kV/70A high voltage switch has been installed at Korea Atomic Energy Research Institute(KAERI) in Taejon to supply power with Korea Superconducting Tokamak Advanced Research(KSTAR) Neutral Beam Injection(NBI) system. NBI system requires fast cutoff of the flower supply voltage for protection of the grid when arc detected and fast turn-on the voltage for sustaining the beam current. Therefore the high voltage switch and arc current detection circuit are important part of the NBI power supply. There are much need for high voltage solid state switches in NBI system and a broad area of applications. This switch consisted of 100 series connected MOSFETs and adopted the proposed simple and reliable gate drive circuit without bias supply. Various results taken during the commissioning phase with a 100kW resistive load and NBI source arc shown. This paper presents the detailed design of 120kV/70A high voltage MOSFETs switch and simple gate drive circuit. Problems with the high voltage switch and gate driver during thefabrication and test and solutions are also presented.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.8
• /
• pp.2555-2564
• /
• 2000

In this paper, a new unsupervised image segmentation algorithm is proposed. To model the contextual information presented in images, the characteristics of the Markov random fields (MRF) are utilized. Textured images are modeled as realizations of the stationary Gaussian MRF on a two-dimensional square lattice using the conditional autoregressive (CAR) equations with a second-order noncausal neighborhood. To detect boundaries, hypothesis tests over two masked areas are performed. Under the hypothesis, masked areas are assumed to belong to the same class of textures and CAR equation parameters are estimated in a minimum-mean-square-error (MMSE) sense. If the hypothesis is rejected, a measure of dissimilarity between two areas is accumulated on the rejected area. This approach produces potential edge maps. Using these maps, boundary detection can be performed, which resulting no micro edges. The performance of the proposed algorithm is evaluated by some experiments using real images as weB as synthetic ones. The experiments demonstrate that the proposed algorithm can produce satisfactorY segmentation without any a priori information.

• Journal of the Korean Geotechnical Society
• /
• v.19 no.6
• /
• pp.189-206
• /
• 2003

During tunnel excavation in urban area a systematic monitoring is important for the purpose of determination of support type and quantity, as well as for the control of stability of both surface structures and the tunnel itself due to the frequently, and in many cases, abruptly changing ground condition. In Austria absolute displacement monitoring methods have replaced relative displacement measurements by geodetic methods to a large extent. Prompt detection of weak ground ahead of the tunnel face as well as better adjustment of excavation and support to the geotechnical conditions is possible with the help of the improved methods of data evaluation on sites. Deformation response of the ground to excavation starts ahead of the tunnel face, therefore, the deformation and state of the tunnel advance core is the key factor of the whole deformation process after excavation. In other words, the rigidity and state of the advance core play a determining role in the stability of both surface structures and the tunnel itself. This paper presents the results from detailed three-dimensional numerical studies, exploring vertical displacements, vector orientations and extrusions on tunnel face during the progressive advancement for the shallow tunnel in various geotechnical conditions.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.22 no.10
• /
• pp.2236-2245
• /
• 1997

The DBS system is being highlighted as actual area for the information societics. Specially, the DBS have been proposed very useful system to access the broading service in more widely sea. But the antenna tracking system for maritime DBS receiving is requried complicated control system because of the those complex motion represented pitching, rolling and yowing etc. Our resesrch target is a development of tracking system to the KOREASEA(MUGUNGWHA-1,2) for the applicated small size shipping. So our development focus was concentrated the two development direction. The first focus was represented low-cost system for popularization of small-size shipping around sea of the Korea peninsula. The second focus was an adaptive possibilities with domestic eqdupiment which was developed satellite receiving for KOREASAT. The anntenna mount is designed a compact size and easy operation use to the Az/El 2-axis type which is operated by step motor. And this mount type is very useful on a ship in the near sea of Korea peninsula. Basic tracking method is used th step-tracking algorithm, and the ship's moving compensation is adapted to the closed loop control method by ship's moving detection of gyro sensor. Control part is consists of converter, countertime, VCO, micro-computer and it's software. Testing the operation by the ship's moving simulator, and algorithm is designed tracking and moving compensation by receiving state.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.33 no.4A
• /
• pp.377-386
• /
• 2008

In this paper, a new handover management scheme has been proposed to reduce handover latency and to support fast handover without packet losses, so that it may be applicable to the wireless mobile Internet system such as IPv6-based WiBro system. To minimize the handover latency in processing of movement detection, we propose the handover management scheme which simplifies the handover message exchanging procedure between mobile subscriber station (MSS) and network by integrating layer 2 and layer 3 handovers efficiently based on the layer2 information. To reduce the processing delay from new care-of-address (NCoA) configuration during handover, we propose that NCoA is created, distributed and managed by new access control router (NACR). In addition, in order to minimize the packet transmission delay and eliminate the packet losses, the proposed scheme employs a crossover router (CR) which is upper network located over PACR and NACR and employs the packet buffering for MSS. The simulation study shows that the proposed scheme achieves loss-free packet delivery and low latency in the environment of narrow overlapped cell area or high velocity of the MSS, comparing the performance with the conventional schemes.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2004.11a
• /
• pp.456-459
• /
• 2004

Analog film/screen systems have been being changed to a digital x-ray imaging device using direct conversion materials. Photocoductors for a direct detection flat-panel imager require high x-ray absorption, ionization and charge collection, low leakage current and large area deposition. In this work, $HgI_2$ films with excellent properties for x-ray detector were deposited by screen printing method. The thickness of $HgI_2$ film was about $150\;{\mu}m$. The passivation layer is fabricated using a-Se and parlyene, the both fabrication $HgI_2$ film were compared for analyzing the leakage current reduction. We measured electrical properties-leakage current, photosensitivity, SNR though I-V measurement, As the result, $HgI_2$ film using a-Se passivation layer had the greater

• Journal of IKEEE
• /
• v.2 no.1 s.2
• /
• pp.34-41
• /
• 1998

An integrated circuit for magnetic stripe storage system is implemented. All the analog and digital circuits are integrated in one chip. The analog block contains preamplifier, peak detecter, comparator and reference generater. And digital block includes reference window signal generater, up/down counter for F/2F signal measurement, bit-error detection logic, and control logic. Both the encoding and decoding functions for F/2F signal processing are provided. An AGC(automatic gain control) circuit which was included in conventional circuits is eliminated due to optimized circuit design. Misreading prevention circuits are also proposed by fixing up new reference bit when broken bits are detected. The prototype chip is implemented using $0.8{\mu}m$ N-well CMOS technology and operates from 3.3 V to 7.5 V of supply voltage. It occupies a die area of $3.04mm^2(1.6mm{\times}1.9mm)$ and dissipates 8 mW with a 5 V supply voltage.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1469-1472
• /
• 2019

CAN bus in automotive applications does not assign node addresses. When a node is hacked and it transmits malicious data frame, it is difficult to resolve which node is hacked. However, this CAN bus internal attack seriously threatens the safety of a car, so a prompt counterattack is necessary in the CAN bus physical layer. This paper proposes a counterattack method against malicious CAN bus internal attack. When a malicious data frame is detected, an intrusion detection system in the CAN bus increases the error counter of the malicious node. Then, the malicious node is off from the bus when its error counter exceeds its limit. A CAN controller with the proposed method is implemented in Verilog HDL, and the proposed method is proved to counterattack against malicious CAN bus internal attack.