• KIPS Transactions on Software and Data Engineering
• /
• v.9 no.7
• /
• pp.213-220
• /
• 2020

The convolutional neural network is composed of convolutional layers and fully connected layers. The nonlinear activation function is used in each layer of the convolutional layer and the fully connected layer. The activation function being used in a neural network is a function that simulates the method of transmitting information in a neuron that can transmit a signal and not send a signal if the input signal is above a certain criterion when transmitting a signal between neurons. The conventional activation function does not have a relationship with the loss function, so the process of finding the optimal solution is slow. In order to improve this, an agile activation function that generalizes the activation function is proposed. The agile activation function can improve the performance of the deep neural network in a way that selects the optimal agile parameter through the learning process using the primary differential coefficient of the loss function for the agile parameter in the backpropagation process. Through the MNIST classification problem, we have identified that agile activation functions have superior performance over conventional activation functions.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.2090-2092
• /
• 2003

Adaptive algorithms are widely used for various applications. One challenging application is an echo canceller in the long distance telephony network. This paper proposes the new variable step-size algorithm for adaptive lattice structure for echo cancellation. The new algorithm is using power of the output signal and the error signal to controlled the step of adaptation process. By this technique, the proposed algorithm is an excellent and effective in good stability. Performance comparison of the proposed algorithm and the other algorithm is made through simulation results.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.11
• /
• pp.2563-2568
• /
• 2010

SAII (synthetic aperture integral imaging) is a useful technique to record many multi view images of 3D objects by using a moving camera and to reconstruct 3D depth images from the recorded multiviews. This is largely composed of two processes. A pickup process provides elemental images of 3D objects and a reconstruction process generates 3D depth images computationally. In this paper, a signal model for SAII is presented. We defined the granular noise and analyzed its characteristics. Our signal model revealed that we could reduce the noise in the reconstructed images and increase the computational speed by reducing the shifting distance of a single camera.

• Journal of information and communication convergence engineering
• /
• v.7 no.4
• /
• pp.507-515
• /
• 2009

In this paper, we propose an image signal processor (ISP) for an ultra small HD-grade video sensor with 3A (AWB, AE, and AF) in camera phones that can process 720P/30fps videos. In order to enhance the video quality of the systems, it is necessary to achieve the high performance of the 3A. The proposed AWB algorithm multiplies the adjusted coefficients of color gains to the captured data of white objects. The proposed AE method adopts the index step moving based on the difference between an averaged Y luminance and a target luminance, together with IIR filters with variable time responses. The proposed AF technique controls the focus curve to find the lens position that maximizes the integrated high frequency components in luminance values by using highpass filters. Finally, we compare the image quality captured from our system to the quality of a commercial HD camcorder in order to evaluate the performance of the proposed ISP. The proposed ISP system is also fabricated with 0.18um CMOS flash memory process.

• Proceedings of the IEEK Conference
• /
• 2006.06a
• /
• pp.1013-1014
• /
• 2006

Mobilecomputer offers more fundamental role than role assistance enemy of modern technology equipment and new Information Technology can reconsider, and reconstruct creatively accuracy of physiological concept. That military register symptoms are developed of disease, before far before rehalibitation, offer possibility that can intervene in process that motive change of military register symptoms after rehalibitation. But, that many parameters become analysis target and mathematical settlement and equalization system of neted data of that is huge, same time collection of all datas can lift difficulty etc.. These main weakness puts in structural relation between elements that compose system. Therefore, dynamics research that time urea of systematic adjustment has selected method code Tuesday nerve dynamics enemy who groping of approach that become analysis point is proper and do with recycling bioelectricity signal. Nature model of do living body signal digital analysis chapter as research result could be developed and scientific foundation groping could apply HSS (Hardware-software system) by rehalibitation purpose. Special quality that is done radish form Tuesday of bioelectricity signal formation furthermore studied, and by the result, fundamental process of bodysignal in do structure circuit form of analog - digital water supply height modelling do can

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39C no.1
• /
• pp.60-69
• /
• 2014

Footstep detection using seismic sensors for security is a very meaningful task, but readings can easily fluctuate due to noise in outdoor environment. We propose NSSC method based on nonlinear spectral subtraction and cross-correlation using prime footstep model signal as a footstep signal refining process that enhances the signal-to-noise ratio (SNR) and attenuates noise. After de-noising, a detection event classification method is presented as further refining process to ensure that the detection result is a footstep. To validate the proposed algorithm, representative experiments including sunny and rainy-day cases are demonstrated.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.25 no.6B
• /
• pp.1127-1135
• /
• 2000

The beamforming procedure in Adaptive Array Antenna System is affected by signal degradation and data rate due to DS-CDMA characteristics. Until this time, a lot of techniques are suggested to overcome this problems. This paper shows the simulation result about the beamforming performance of symbol level system that process slow data rate, compensated signal by despreading procedure in front of beamformer, and that of chip level system that process chip level signal without it. we analysis the performance using MSE, beam pattern, scattering points of beamformer output. chip level system is superior to symbol level system in time varying channel, while the performance of them didn't have difference in static channel.

• Transactions of the Society of Information Storage Systems
• /
• v.3 no.4
• /
• pp.160-166
• /
• 2007

In this paper, the measured electrical resistance method is proposed to analyze the ramp-tab contact during the load/unload (L/UL) process. Since this method supplies the voltage change due to the resistance change, we can easily and conveniently identify the ramp-tab contact from the acoustic emission (AE) signal. At first, we carefully deposit the conductive material on the surface of the conventional ramp by sputtering method. The ratio frequency (RF) magnetron co-sputtering system is applied to accomplish the deposited double-layers on the ramp surface. One layer is the stainless steel for the conductive layer and the other is the titanium layer for the cohesive function between the ramp surface and the stainless steel layer. In order to guarantee the stiffness and damping properties of the original ramp, the deposited conductive layer is intended to have very thin thickness. After integration the proposed ramp device into the L/UL system and networking the electrical resistance circuit, the L/UL performance is experimentally evaluated by comparing the measured electrical resistance signal and AE signal.

• Journal of Auto-vehicle Safety Association
• /
• v.14 no.2
• /
• pp.70-74
• /
• 2022

This paper describes the development of a target vehicle state estimation algorithm using vehicle-to-vehicle (V2V) communication. Perceiving the state of the target vehicle has great importance for successful autonomous driving and has been studied using various sensors and methods for many years. V2V communication has advantage of not being constrained by surrounding circumstances relative to other sensors. In this paper, we adopt the V2V signal for estimating the target vehicle state. Since applying only the V2V signal is improper by its low frequency and latency, the signal is used as additional measured data to improve the estimation accuracy. We estimate the target vehicle state using Extended Kalman filter (EKF); a point mass model was utilized in process update to predict the state of next step. The process update is followed by measurement update when ego vehicle receives V2V information. The proposed study evaluated state estimation by comparing input V2V information in an experiment where the ego vehicle follows the target vehicle behind it.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.11
• /
• pp.2685-2690
• /
• 2014

WPS determines position via a two-step process. In the construction phase, a radio map is constituted by collecting radio information signals. Positioning procedure is a step of comparing the radio signal newly acquired with the radio map. If the signals collected and the radio map are different the accuracy decreases. Even though the rate of accuracy is different according to positioning algorithms, accuracy drop is an issue common to all WPS systems. Signal strength fluctuation is caused by the malfunction of the device that receives positioning signals, obstruction and channel interference, etc. In this paper, in order to solve the problem caused by signal strength change, we propose a new radio map construction technique. The proposed method is intended to constitute a strong radio map to changes in the signal strength and updated by collecting the signal strength changes to the radio map. The use of this method is expected to enhance the accuracy of WPS by actively counteracting signal fluctuation.