• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.5
• /
• pp.259-265
• /
• 2023

Pedestrian detection is used to effectively detect pedestrians in various situations based on deep learning. Pedestrian detection has difficulty detecting pedestrians due to problems such as camera performance, pedestrian description, height, and occlusion. Even in the same pedestrian, performance in detecting them can differ according to the height of the pedestrian. The height of general pedestrians encompasses various scales, such as those of infants, adolescents, and adults, so when the model is applied to one group, the extraction of data becomes inaccurate. Therefore, this study proposed a pedestrian detection method that fine-tunes the pedestrian area by Refining Layer and Feature Concatenation to consider various heights of pedestrians. Through this, the score and location value for the pedestrian area were finely adjusted. Experiments on four types of test data demonstrate that the proposed model achieves 2-5% higher average precision (AP) compared to Faster R-CNN and DRPN.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.11 no.3
• /
• pp.163-171
• /
• 2016

Lane detection is a widely researched topic. Although simple road detection is easily achieved by previous methods, lane detection becomes very difficult in several complex cases involving noisy edges. To address this, we use a Convolution neural network (CNN) for image enhancement. CNN is a deep learning method that has been very successfully applied in object detection and recognition. In this paper, we introduce a robust lane detection method based on a CNN combined with random sample consensus (RANSAC) algorithm. Initially, we calculate edges in an image using a hat shaped kernel, then we detect lanes using the CNN combined with the RANSAC. In the training process of the CNN, input data consists of edge images and target data is images that have real white color lanes on an otherwise black background. The CNN structure consists of 8 layers with 3 convolutional layers, 2 subsampling layers and multi-layer perceptron (MLP) of 3 fully-connected layers. Convolutional and subsampling layers are hierarchically arranged to form a deep structure. Our proposed lane detection algorithm successfully eliminates noise lines and was found to perform better than other formal line detection algorithms such as RANSAC

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.06a
• /
• pp.762-766
• /
• 1998

Mechanical valve is one of the most widely used implantable artificial organs, Since its failure (mechanical failures and thrombosis to name two representative example) means the death of patient, its reliability is very important and early noninvasive detection is essential requirement . This paper will explain the method to detect the thrombosis formation by spectral analysis and neural network. In order quantitatively to distinguish peak of a normal valve from that of a thrombotic valve, a 3 layer backpropagation neural network, which contains 7,000 input nodes, 20 hidden layer and 1output , was employed. The trained neural network can distinguish normal and thrombotic valve with a probability that is higher than 90% . In conclusion, the acoustical spectrum analysis coupled with a neural network algorithm lent itself to the noninvasive monitoring of implanted mechanical valves. This method will be applied to be applied to the performance evaluation of other implantable rtificial organs.

• Journal of Sensor Science and Technology
• /
• v.20 no.6
• /
• pp.393-399
• /
• 2011

In this paper, we designed and fabricated the novel air flow sensor using air drag force, which can be applied to the low air flow detection. To measure the low air flow, we should enlarge the air drag force and the output signal at the given air flow. The paddle structure is applied to the device, and the device is vertically located against the air flow to magnify the air drag force. We also adapt the corrugation structure to improve the output signals on the given air velocity. The device structure is made up of the silicon nitride layer and the output signal is measured with the piezoresistive layer. The output signals from the corrugated device show the better measurement sensitivity and the response time than that of flat one. The repeated measurement also shows the stabilized signals.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 1994.11a
• /
• pp.118-121
• /
• 1994

The NO$_2$ gas-detection characteristics were investigated using the functional organic Langmuir-Blodgett(LB) film of Copper-tetra-tert-butylphalocyanine(CUTBP). The optimum conditions for a film deposition were obtained through a study of $\pi$-A isotherm and the deposited film status was confirmed by ellipsometry measurements. In the NO$_2$ response experiments, first of all, the proper number of layer was obtained by a change of the electrical conductivities depending on the number of layer when the film is exposed to the gas. And response time, recovery time, and reproducibility were measured on it. On the other hand, how the NO$_2$ gas interacts to the LB film was studied by UV/visible absorption spectra.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2010.07a
• /
• pp.525-526
• /
• 2010

2000년 대규모 DDoS 공격이래, 2009년 7월 7일 국가주요정부기관 및 인터넷 포털, 금융권 등의 웹사이트 대상으로 1차, 2차, 3차로 나누어 대규모 사이버 공격이 발생하였다. 지속적으로 발전되는 행태를 보이고 DDoS 공격에 대해 본 논문에서는 계층적인 침입탐지시스템을 설계하였다. 네트워크 패킷을 분석하기 위해 e-Watch, NetworkMiner등의 패킷, 프로토콜 분석도구를 이용하여 TCP/IP의 Layer별 공격을 분석한 후 패킷의 유입량, 로그정보, 접속정보, Port, Address 정보를 분석하고 계층침입에 대한 방어를 수행하도록 설계하였다. 본 논문은 DDoS(Distributed Denial of Service)에 대한 패킷 전송에 대해 계층적인 방어를 통해 보다 안정적인 패킷수신이 이루어진다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.4A
• /
• pp.241-249
• /
• 2005

ATM based MPLS(Multiprotocol Label Switching) is discussed for its provisioning QOS commitment capabilities, Traffic engineering and smooth migration for BcN in Korea. At this time, due to the comprehensive nature of ATM protocol, ATM has been adapted as the backbone system for carrying Internet traffic[1,2,3,4]. This paper presents preventive congestion control mechanisms for detecting HTR(Hard-To-Reach) LSP(Label Switched Path) in ATM based MPLS systems. In particular, we have introduced a HTR LSP detection method using network signaling information in an ATM layer. MPLS related studies can cover LSP failures in a physical layer fault, it can not impact network congestion status. Here we will present the research results for introducing HTR LSP detection methods and control mechanisms and this mechanism can be implementing as SOC for high speed processing a packet header. We concluded that it showed faster congestion avoidance abilities with a more reduced system load and maximized the efficiency of network resources by restricting ineffective machine attempts.

• The Transactions of the Korean Institute of Electrical Engineers C
• /
• v.50 no.9
• /
• pp.444-449
• /
• 2001

In this paper, a novel ultraviolet sensor is presented based on a photoluminescent porous silicon. Porous silicon layer was formed by chemical etching of surface of pn junction in a $HF(48%)-HNO_3(60%)-H_20$ solution. Incident ultraviolet(UV) light is converted to visible light by photoluminescent porous silicon layer, and then this visible light generates electron-hole pairs in the pn junction, which produces a photocurrent flow through the device. In order to maximize detection efficiency, the peak sensitivity wavelength of the pn junction diode was matched with the peak wavelength of Photoluminescence from porous silicon layer. The porous silicon ultraviolet sensor showed a large output current as UV intensity increases and but very low sensitivity to visible light. The detection sensitivity of porous silicon sensor was calculated as 2.91mA/mW. These results are expected to open up a possibility that the present porous silicon sensor can be used for detecting UV light in a visible background, compared to silicon UV detectors which have an undesirable response to visible light.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.13 no.2
• /
• pp.171-181
• /
• 1988

In this study, a packet voice/data terminal(PVDT) that services both voice and data in the packet-switched network is implemented. The software structure of the PVDT is designed according to the OSI 7 layer architecture. The discrimination of voice and data is made in the link layer. Voice packets have priority over data packets in order to minimize the transmission delay, and are serviced by a simple protocol so that the overhead arising form the retransmission of packets may be minimized. The hardware structure of the PVDT is divided into five modules; a master control module, a speech proessing module, a speech activity detection module, a telephone interface module, and an input/output interface module. In addition to the hardware implementation, the optimal reconstruction delay of voice packets to reduce the influence of delay variance is analyzed.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.6 no.9
• /
• pp.2323-2340
• /
• 2012

Recent advances in cognitive radio technology have drawn immense attention to higher layer protocols above medium access control, such as transmission control protocol (TCP). Most proposals to improve the TCP performance in cognitive radio (CR) networks have assumed that either all nodes are in CR networks or the TCP sender side is in CR links. In those proposals, lower layer information such as the CR link status could be easily exploited to adjust the congestion window and improve throughput. In this paper, we consider a TCP network in which the TCP sender is located remotely over the Internet while the TCP receiver is connected by a CR link. This topology is more realistic than the earlier proposals, but the lower layer information cannot be exploited. Under this assumption, we propose an enhanced TCP protocol for CR networks called TCP for cognitive radio (TCP-CR) to improve the existing TCP by (1) detection of primary user (PU) interference by a remote sender without support from lower layers, (2) delayed congestion control (DCC) based on PU detection when the retransmission timeout (RTO) expires, and (3) exploitation of two separate scales of the congestion window adapted for PU activity. Performance evaluation demonstrated that the proposed TCP-CR achieves up to 255% improvement of the end-to-end throughput. Furthermore, we verified that the proposed TCP does not deteriorate the fairness of existing TCP flows and does not cause congestions.