• The KIPS Transactions:PartB
• /
• v.13B no.4 s.107
• /
• pp.489-498
• /
• 2006

Clustering results differ according to the datasets and the performance worsens even while using web documents which are manually processed by an indexer, because although representative clusters for a feature can be obtained by statistical feature selection methods, irrelevant features(i.e., non-obvious features and those appearing in general documents) are not eliminated. Those irrelevant features should be eliminated for improving clustering performance. Therefore, this paper proposes three feature-filtering algorithms which consider feature values per document set, together with distribution, frequency, and weights of features per document set: (l) features filtering algorithm in a document (FFID), (2) features filtering algorithm in a document matrix (FFIM), and (3) a hybrid method combining both FFID and FFIM (HFF). We have tested the clustering performance by feature selection using term frequency and expand co link information, and by feature filtering using the above methods FFID, FFIM, HFF methods. According to the results of our experiments, HFF had the best performance, whereas FFIM performed better than FFID.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.3E
• /
• pp.110-112
• /
• 2003

There have been numerous studies on the enhancement of the noisy speech signal. In this paper, we propose a completely new speech enhancement scheme, that is, a filtering of a dissonant frequency (especially F# in each octave of the tempered scale) based on the fundamental frequency which is developed in frequency domain. In order to evaluate the performance of the proposed enhancement scheme, subjective tests (MOS tests) were conducted. The subjective test results indicate that the proposed method provides a significant gain in audible improvement especially for speech contaminated by colored noise and speaking in a husky voice. Therefore when the filter is employed as a pre-filter for speech enhancement, the output speech quality and intelligibility is greatly enhanced.

• 전자공학회논문지 IE
• /
• v.45 no.1
• /
• pp.24-30
• /
• 2008

In this paper, Digital MultiTone(DMT) is an emerging multi-carrier modulation scheme, which has been adopted for VDSL(Very high speed Digital Subscribe Line). A problem of DMT is its sensitivity to frequency offset between the transmitted and received carrier frequencies. This frequency offset introduces inter-carrier interference(ICI) in the DMT symbol. This paper is proposed an ICI cancelation scheme using Kalman Filtering. The performance of the proposed method is compared with conventional methods in terms of bit error rate performance, bandwidth efficiency, and computational complexity. Through simulations, it is shown that for high values of the frequency offset and for higher order modulation schemes, the EKF(Enhanced Kalman Filtering) method perform better than the others.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2004.10a
• /
• pp.447-449
• /
• 2004

The purpose of this study is measured surface acoustic wave(SAW) characteristics to confirm utilization possibility as SAW sensor using new Pb(Mg$_{1}$3/Nb$_{2}$3/) $O_3$-PbTiO$_3$ (PMN-PT) piezoelectric substrate. We have tried to see if the material can be practically available as a new surface acoustic wave (SAW) biosensor to detect protein. The experimental results clarified that the frequency filtering of the central frequency of the PMN-PT substrate is a superior result to that of the LiTaO$_3$ (LT) substrate, but the result was not completely satisfactory. We know there is a problem in the design of inter-digital transducer (IDT) pattern. The waves transferred through the input terminal forms SAW which is sure to be transferred to the direction of the output terminal and the backward direction of the input terminal. This reflected wave is reiterated with SAW, which is transferred to the output direction, and so the frequency filtering gives a not good result. The electromechanical coupling coefficient of the PMN-PT substrate is excellent, and we can use it as a SAW sensor, in the near future, provided that there will be a new IDT design to increase the frequency filtering.

• ETRI Journal
• /
• v.37 no.1
• /
• pp.61-65
• /
• 2015

In this paper, a novel filtering power divider with external isolation resistors is presented. The proposed power divider can be considered as an integration of a bandpass filter and a Gysel power divider. Based on the circuit topology, a high-order filtering power divider can be easily realized. Odd- and even-mode models are employed to analyze the filtering and power splitting functions. For demonstration, a third-order filtering power divider operating at 1.5 GHz is designed and implemented. The measured results exhibit an isolation between the output ports that is better than 20 dB at around the center frequency.

• Proceedings of the KSRS Conference
• /
• 2004.10a
• /
• pp.237-240
• /
• 2004

In this study, the digital image processing with image enhancement based on homomorphic filtering was performed using geophysical imaging data such as gravity, magnetic data and sub-scenes of satellite images such as LANDSAT, IKONOS, and KOMPSAT. Windows application program for executing homomorphic filtering was designed and newly implemented. In general, homomorphic filtering is technique that is based on Fourier transform, which enhances the contrast of image by removing the low frequencies and amplifying the high frequencies in frequency domain. We can enhance the image selectively using homomorphic filtering as compared with the existing method, which enhance the image totally. Through several experiment using remotely sensed imagery and geophysical image with this program, it is concluded that homomorphic filtering is more effective to reveal distinct characteristics for some complicated and multi-associated features on image data.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.22 no.4
• /
• pp.706-713
• /
• 1997

In this paper, we propose a band stop filter(BSF) for reducing drag-like effect of the low pass filter(LPF), a block by block adaptive filtering method, and a motion adaptive filtering method, which show better results in terms of PSNR or human visual perception compared to the conventional method using LPF. The BSF improves the draglike effects of the low pass filter by passing temporal high frequency components of video sequences which correspond to objects with large motion. The proposed adaptive methods also improve the conventional adaptive filtering by modifying the conventional algorithm and applying the algorithms for small blocks. The simulation results show that the proposed filtering methods show better results in terms of PSNR and subjective tests in most cases. Also in case of block by block adaptive filtering, it is verified that the application of the algorithm for smaller block gives better results.

• MALSORI
• /
• no.60
• /
• pp.181-190
• /
• 2006

This paper describes a speech enhancement scheme that leads to significant improvements in recognition performance when used in the ASR front-end. The proposed approach is based on adaptive comb filtering and an MMSE-related parameter estimator. While adaptive comb filtering reduces noise components remarkably, it is rarely effective in reducing non-stationary noises. Furthermore, due to the uniformly distributed frequency response of the comb-filter, it can cause serious distortion to clean speech signals. This paper proposes an improved comb-filter that adjusts its spectral magnitude to the original speech, based on the speech absence probability and the gain modification function. In addition, we introduce the modified comb filtering-based speech enhancement scheme for ASR in mobile environments. Evaluation experiments carried out using the Aurora 2 database demonstrate that the proposed method outperforms conventional adaptive comb filtering techniques in both clean and noisy environments.

• Korean Journal of Optics and Photonics
• /
• v.30 no.2
• /
• pp.48-58
• /
• 2019

We investigate the signal-to-noise ratios (SNRs) of time-domain (i.e. pulsed illumination) and frequency-domain (i.e. chirped illumination) photoacoustic signals measured by a spherically focused ultrasound transducer for spherical absorbers. The simulation results show that the time-domain photoacoustic SNR is higher than that of frequency-domain photoacoustic signals, as reported in the previous literature. We understand the reason for this SNR gap between the two measurement modes by analyzing photoacoustic-signal spectra, considering the incident beam energy controlled by the maximum permissible exposure. As the result of this approach, we find that filtering off the DC term in the chirped signal's spectrum improves frequency-domain photoacoustic SNRs by up to approximately 5 dB. In particular, it is observed that photoacoustic SNRs are highly sensitive to an upper-frequency value of frequency filtering functions, and the optimal upper-frequency values maximizing the SNR are different in time- and frequency-domain photoacoustic measurements.

• Journal of IKEEE
• /
• v.23 no.1
• /
• pp.310-313
• /
• 2019

This paper used the majority function to digitally filter back-electromotive force as an explanation of the Brushless DC MOTOR control algorithm. The cause and improvement of motor noise, which are operating in close proximity to high frequency sources, did not use conventional low pass filter and comparator elements. Also, they repeatedly output a noise-free BEMF signal for the input value of the majority detection filtering. These filtering steps can help reduce costs and minimize the area of a PCB by requiring relatively little hardware.