• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.1
• /
• pp.50-56
• /
• 2020

With the rapid development of deep neural network theory and application research, the effect of face detection has been improved. However, due to the complexity of deep neural network calculation and the high complexity of the detection environment, how to detect face quickly and accurately becomes the main problem. This paper is based on the relatively simple model of the MTCNN model, using FDDB (Face Detection Dataset and Benchmark Homepage), LFW (Field Label Face) and FaceScrub public datasets as training samples. At the same time of sorting out and introducing MTCNN(Multi-Task Cascaded Convolutional Neural Network) model, it explores how to improve training speed and Increase performance at the same time. In this paper, the dynamic image pyramid technology is used to replace the traditional image pyramid technology to segment samples, and OHEM (the online hard example mine) function in MTCNN model is deleted in training, so as to improve the training speed.

• Journal of the Korean Society of Industry Convergence
• /
• v.26 no.5
• /
• pp.953-965
• /
• 2023

The stratospheric drones are developed to perform missions such as weather observation, communication relay, surveillance, and reconnaissance at 18km to 20km, where climate change is minimal and there is no worry about a collision with aircraft. It uses solar panels for daytime flights and energy stored in batteries for night flights, providing many advantages over existing satellites. The electrical and power systems essential for stratospheric drone flight must ensure reliability, efficiency, and lightness by selecting the optimal circuit topology. Therefore, it is necessary to analyze the circuit topology of various types of multi-level inverters with high redundancy that can ensure the reliability and efficiency of the motor driving power required for stable long-term flight of stratospheric drones. By quantifying the switch element voltage drop and the number and weight of inverter components for each topology, we evaluate efficiency and lightness and propose the most suitable circuit topology for stratospheric drones.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.19 no.1
• /
• pp.301-307
• /
• 2024

This study introduces a novel cascade fusion architecture aimed at enhancing thermal images across various scale conditions. The processing of thermal images at multiple scales has been challenging due to the limitations of existing methods that are designed for specific scales. To overcome these limitations, this paper proposes a unified framework that utilizes cascade feature fusion to effectively learn multi-scale representations. Confidence maps from different image scales are fused in a cascaded manner, enabling scale-invariant learning. The architecture comprises end-to-end trained convolutional neural networks to enhance image quality by reinforcing mutual scale dependencies. Experimental results indicate that the proposed technique outperforms existing methods in multi-scale thermal image enhancement. Performance evaluation results are provided, demonstrating consistent improvements in image quality metrics. The cascade fusion design facilitates robust generalization across scales and efficient learning of cross-scale representations.

• Korean Journal of Agricultural Science
• /
• v.36 no.1
• /
• pp.87-98
• /
• 2009

The objective of this study is to evaluate the effect of increasing instream flow at Gapcheon stream of Daejeon city by considering two virtual reservoirs upstream, respectively; Geum-gok reservoir and Koe-gok reservoir upstream, respectively. The paralleled and cascaded reservoir operations were performed including the existing Jang-an and Bang-dong reservoirs. The results are summarized as follows. Firstly, from the Bang-dong and Geum-gok cascaded reservoir's water balance analysis, instream flow of $6.83Mm^3$ was able to be supplied to downstream, and water supply indexes of Geum-gok reservoir were analyzed to have the rate of water supply divided by watershed area of 403.4 mm, the rate of water supply divided by rainfall of 33.0 %, the rate of water supply divided by inflow of 96.4 %, the rate of water supply divided by storage capacity of 81.9 %, and the rate of inflow divided by storage capacity of 112.3 %. Secondly, from the Jang-an and Geum-gok paralleled reservoir's water balance analysis, flow durations at Gapcheon station were analyzed to have Q95 (the 95th high flow) of $4.806m^3/s$, Q185 (the 185th high flow) of $2.217m^3/s$, Q275 (the 275th high flow) of $1.140m^3/s$, and Q355 (the 355th high flow) of $0.887m^3/s$. Thirdly, inflow to Koe-gok reservoir was simulated including the Jang-an and Bang-dong paralleled reservoir's water balance analysis, instream flow of $49.60Mm^3$ was able to be supplied from Koe-gok reservoir to downstream, and water supply indexes of Koe-gok reservoir were analyzed to have the rate of water supply divided by watershed area of 246.5 mm, the rate of water supply divided by rainfall of 19.4 %, the rate of water supply divided by inflow of 40.8 %, the rate of water supply divided by storage capacity of 412.1 %, and the rate of inflow divided by storage capacity of 1,189.8 %. Fourthly, daily streamflows at Gapcheon stream were simulated including outflows from Koe-gok reservoir, flow durations at Gapcheon station were analyzed to have Q95 (the 95th high flow) of $4.501m^3/s$, Q185 (the 185th high flow) of $2.277m^3/s$, Q275 (the 275th high flow) of $1.743m^3/s$, and Q355 (the 355th high flow) of $1.564m^3/s$. The conclusion appeared that the effect of increasing instream flow at Gapcheon stream from Koe-gok reservoir was more higher than that from Geum-gok reservoir.

• Korean Journal of Agricultural Science
• /
• v.35 no.2
• /
• pp.225-235
• /
• 2008

The objective of this study is to evaluate the hydrological feasibility of heightening the Dae-ah reservoir in order to save instream flow at the Bong-dong station situated in the Mankyoung river. The results are summarized as follows. Firstly, from the Dong-sang and Dae-ah cascaded reservoir's water balance analysis, water supply indexes of the Dae-ah reservoir were analyzed to have the rate of water supply divided by watershed area of 1207.4 mm, the rate of water supply divided by rainfall of 95.8%, the rate of water supply divided by inflow of 153.1%, the rate of water supply divided by storage capacity of 236.1%, and the rate of inflow divided by storage capacity of 200.6%. Secondly, from the Dae-ah and Kyoung-cheon paralleled reservoir's water balance analysis, flow durations at the Bong-dong station were analyzed to have the Q95 (the 95th high flow) of $28.95m^3/s$, the Q185 (the 185th high flow) of $2.00m^3/s$, the Q275 (the 275th high flow) of $2.00m^3/s$, and the Q355 (the 355th high flow) of $0.82m^3/s$. Thirdly, in case of heightening the full water level of the Dae-ah reservoir of 10m, from the Dong-sang and Dae-ah cascaded reservoir's water balance analysis, water supply indexes of the Dae-ah reservoir were analyzed to have the rate of water supply divided by watershed area of 1220.7 mm, the rate of water supply divided by rainfall of 96.8%, the rate of water supply divided by inflow of 154.6%, the rate of water supply divided by storage capacity of 160.0%, and the rate of inflow divided by storage capacity of 137.0%. Fourthly, in case of heightening the full water level of the Dae-ah reservoir of 10m, from the Dae-ah and Kyoung-cheon paralleled reservoir's water balance analysis, flow durations at the Bong-dong station were analyzed to have the Q95 of $28.09m^3/s$, the Q185 of $1.79m^3/s$, the Q275 of $1.79m^3/s$, and the Q355 of $0.82m^3/s$. The conclusion appeared not to have the hydrological feasibility of heightening the Dae-ah reservoir from the reason that increased storage capacity does not increase water supply amount any more because of the high rate of the water supply divided by inflow.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.49 no.8
• /
• pp.22-26
• /
• 2012

This paper introduces a 2.5-Gb/s optical receiver implemented in a standard 1P4M 0.18um CMOS technology for the applications of active optical HDMI cables. The optical receiver consists of a differential transimpedance amplifier(TIA), a five-stage differential limiting amplifier(LA), and an output buffer. The TIA exploits the inverter input configuration with a resistive feedback for low noise and power consumption. It is cascaded by an additional differential amplifier and a DC-balanced buffer to facilitate the following LA design. The LA consists of five gain cells, an output buffer, and an offset cancellation circuit. The proposed optical receiver demonstrates $91dB{\Omega}$ transimpedance gain, 1.55 GHz bandwidth even with the large photodiode capacitance of 320 fF, 16 pA/sqrt(Hz) average noise current spectral density within the bandwidth (corresponding to the optical sensitivity of -21.6 dBm for $10^{-12}$ BER), and 40 mW power dissipation from a single 1.8-V supply. Test chips occupy the area of $1.35{\times}2.46mm^2$ including pads. The optically measured eye-diagrams confirms wide and clear eye-openings for 2.5-Gb/s operations.

• Journal of Internet Computing and Services
• /
• v.13 no.2
• /
• pp.41-49
• /
• 2012

Wireless network support and extended mobile network environment with exponential growth of smart phone users allow us to utilize the network anytime or anywhere. Malicious attacks such as distributed DOS, internet worm, e-mail virus and so on through high-speed networks increase and the number of patterns is dramatically increasing accordingly by increasing network traffic due to this internet technology development. To detect the patterns in intrusion detection systems, an existing research proposed an efficient algorithm called the jumping window algorithm and analyzed approximately 2,000 patterns in Snort 2.1.0, the most famous intrusion detection system. using the algorithm. However, it is inappropriate from the number of TCAM lookups and TCAM memory efficiency to use the result proposed in the research in current environment (Snort 2.9.0) that has longer patterns and a lot of patterns because the jumping window algorithm is affected by the number of patterns and pattern length. In this paper, we simulate the number of TCAM lookups and the required TCAM size in the jumping window with approximately 8,100 patterns from Snort-2.9.0 rules, and then analyse the simulation result. While Snort 2.1.0 requires 16-byte window and 9Mb TCAM size to show the most effective performance as proposed in the previous research, in this paper we suggest 16-byte window and 4 18Mb-TCAMs which are cascaded in Snort 2.9.0 environment.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.24 no.11
• /
• pp.1055-1063
• /
• 2013

Low noise amplifiers(LNAs) are an integral component of RF receivers and are frequently required to operate at wide frequency bands for various wireless systems. For wideband operation, important performance metrics such as voltage gain, return loss, noise figures and linearity have been carefully investigated and characterized for the proposed LNA. An inductive shunt feedback configuration is successfully employed in the input stage of the proposed LNA which incorporates cascaded networks with a peaking inductor in the buffer stage. Design equations for obtaining low and high input matching frequencies are easily derived, leading to a relatively simple method for circuit implementation. Careful theoretical analysis explains that poles and zeros are characterized and utilized for realizing the wideband response. Linearity is significantly improved because the inductor between gate and drain decreases the third-order harmonics at the output. Fabricated in $0.18{\mu}m$ CMOS process, the chip area of this LNA is $0.202mm^2$, including pads. Measurement results illustrate that input return loss shows less than -7 dB, voltage gain greater than 8 dB, and a little high noise figure around 7~8 dB over 1.5~13 GHz. In addition, good linearity(IIP3) of 2.5 dBm is achieved at 8 GHz and 14 mA of current is consumed from a 1.8 V supply.

• Progress in Medical Physics
• /
• v.21 no.3
• /
• pp.261-273
• /
• 2010

For the use of Indirect-conversion CMOS (complementary metal-oxide-semiconductor) detectors for digital x-ray radiography and their better designs, we have theoretically evaluated the spatial-frequency-dependent detective quantum efficiency (DQE) using the cascaded linear-systems transfer theory. In order to validate the developed model, the DQE was experimentally determined by the measured modulation-transfer function (MTF) and noise-power spectrum, and the estimated incident x-ray fluence under the mammography beam quality of W/Al. From the comparison between the theoretical and experimental DQEs, the overall tendencies were well agreed. Based on the developed model, we have investigated the DQEs values with respect to various design parameters of the CMOS x-ray detector such as phosphor quantum efficiency, Swank noise, photodiode quantum efficiency and the MTF of various scintillator screens. This theoretical approach is very useful tool for the understanding of the developed imaging systems as well as helpful for the better design or optimization for new development.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.19 no.4
• /
• pp.93-102
• /
• 2018

This paper reports a Surge Protective Device (SPD) that is used to protect an automatic metering interface (AMI) power supplies of communication equipment on a low-voltage distribution system from a lightning current. The surge protective device (SPD) can be classified as one-port SPDs and two-port SPDs with decoupling elements depending on the connection type. The protection of internal systems against the lightning current may require a systematic approach consisting of coordinated SPDs. To deal with this, the definition of a lightning protection zone (LPZ) was studied and interpreted through a theoretical review. Because the lightning current resulting from a lightning surge is considerably high, there is limited protection from one SPD; therefore, coordinated cascaded MOV-based SPDs are installed to solve this problem. Regarding the power grid mentioned in this paper, a class II SPD for the low-voltage distribution system installed on the border of LPZ1 and LPZ2, which establish a protection coordination with the Arrester (LA, SA) that corresponds to the LPZO installed on the MOF stage connected to one system were designed to protect various communication (control) equipment, including the automatic meter reading system inside the branch-type electric supply panel of a building, not the incoming side of one system. In addition, performance-related tests were done by a comparison with the existing method through testing, and the optimal design was achieved for the 1-port SPD that uses a series connection and can bleed load current without any decoupling element.