• Journal of Intelligence and Information Systems
• /
• v.5 no.1
• /
• pp.103-123
• /
• 1999

Input filtering as a preprocessing method is so much crucial to get good performance in time series forecasting. There are a few preprocessing methods (i.e. ARMA outputs as time domain filters, and Fourier transform or wavelet transform as time-frequency domain filters) for handling time series. Specially, the time-frequency domain filters describe the fractal structure of financial markets better than the time domain filters due to theoretically additional frequency information. Therefore, we, first of all, try to describe and analyze specially some issues on the effectiveness of different filtering methods from viewpoint of the performance of a neural network based forecasting. And then we discuss about neural network model architecture issues, for example, what type of neural network learning architecture is selected for our time series forecasting, and what input size should be applied to a model. In this study an input selection problem is limited to a size selection of the lagged input variables. To solve this problem, we simulate on analyzing and comparing a few neural networks having different model architecture and also use an embedding dimension measure as chaotic time series analysis or nonlinear dynamic analysis to reduce the dimensionality (i.e. the size of time delayed input variables) of the models. Throughout our study, experiments for integration methods of joint time-frequency analysis and neural network techniques are applied to a case study of daily Korean won / U. S dollar exchange returns and finally we suggest an integration framework for future research from our experimental results.

• Journal of Korea Multimedia Society
• /
• v.16 no.12
• /
• pp.1385-1392
• /
• 2013

There is a variety of researches about recognition system using biometric data these days. In this study, we propose a new algorithm, uses simultaneous equation that made of the edge of objects, to segment an iris region without threshold values from an anterior eye image. The algorithm attempts to find a center area through calculated outskirts information of an iris, and decides the area where the most points are accumulated. To verify the proposed algorithm, we conducted comparative experiments to Hough transform and Daugman's method, based on 50 images anterior eye images. It was found that proposed algorithm is 5 and 75 times faster than on each algorithm, and showed high accuracy of detecting a center point (95.36%) more than Hough transform (92.43%). In foreseeable future, this study is expected to useful application in diverse department of human's life, such as, identification system using an iris, diagnosis a disease using an anterior image.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.9C
• /
• pp.860-870
• /
• 2005

In this paper we propose a new method to derive GRM(Generalized Reed-Muller) coefacients for each $2^{n}$ polarities using cell of karnaugh map(k-map). Generally, there are the serial and parallel method to derive GRM coefficients. As a serial method, Green method generates GRM coefncients using transform matrix. And as a parallel method, Besslich algorithm produces GRM coefficients of each polarity using the generated anteriorly. Green's method generates GRM coefficients for n-variable by calculating transform matrix for one-variable and n-times kronecker product this matrix. And Besslich's method generates GRM coefficients of each polarity in order of Grey-code. But those methods have disadvantages that the number of variable exceeding four makes transform matrix large and there are so many operation steps. In this paper, GRM coefficients is generated by producing cell [$f_{i}$] minimizing variable on k-map and operating this cell [$f_{i}$] and transform matrix for one-variable. So, we can generate GRM coefficients of all polarities easily by using the proposed method.

• Journal of Broadcast Engineering
• /
• v.17 no.1
• /
• pp.26-36
• /
• 2012

Since shape of an object in an image carries important information in contents based image retrieval (CBIR), many shape description methods have been proposed to retrieve images using shape information. Among the existing shape based image retrieval methods, the method which employs invariant Zernike moment desciptor (IZMD) showed better performance compared to other methods which employ traditional Zernike moments descriptor in CBIR. In this paper, we propose a new image retrieval method which applies invariant angular radial transform descriptor (IARTD) to obtain higher performance than the method which employs IZMD in CBIR. IARTD is a rotationally invariant feature which consists of magnitudes and alligned phases of angular radial transform coefficients. To produce rotationally invariant phase coefficients, a phase correction scheme is performed while extracting the IARTD. The distance between two IARTDs is defined by combining the differences of the magnitudes and the aligned phases. Through the experiment using MPEG-7 shape dataset, the average bull's eye performance (BEP) of the proposed method is 0.5806 while the average BEPs of the exsiting methods which employ IZMD and traditional ART are 0.4234 and 0.3574, respectively.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.8 no.3
• /
• pp.570-576
• /
• 2004

In this paper, we design the IFFT/FFT (Inverse fast Fourier Transform/Fast Fourier Transform) modules for IEEE 802.11a-1999, which is a standard of the High-speed Wireless LAN using the OFDM (Orthogonal Frequency Division Multiplexing). The designed IFFT/FFT is the 64-point FFT to be compatible with IEEE 802.11a and the pipelined architecture which needs neither serial-to-parallel nor parallel-to-serial converter. We compare four types of IFFT/FFT modules for the hardware complexity and operation : R22SDF (Radix-2 Single-path Delay feedback), the R2SDF (Radix-2 Single-path Delay feedback), R2SDF (Radix-4 Single-path Delay Feedback), and R4SDC (Radix-4 Single-path Delay Commutator). In order to minimize the error, we design the IFFT/FFT module to operate with additional decimal parts after butterfly operation. In case of the R22SDF, the IFFT/FFT module has 44,747 gate counts excluding RAMs and the minimized error rate as compared with other types. And we know that the R22SDF has a small hardware structure as compared with other types.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.20 no.6
• /
• pp.501-510
• /
• 2000

Recently, advanced signal analysis which is called "time-frequency analysis" has been used widely in nondestructive evaluation applications. Wavelet transform(WT) and Wigner Distribution are the most advanced techniques for processing signals with time-varying spectra. Wavelet analysis method is an attractive technique for evaluation of material characterization nondestructively. Wavelet transform is applied to the time-frequency analysis of ultrasonic echo waveform obtained by an ultrasonic pulse-echo technique. In this study, the feasibility of noise suppression of ultrasonic flaw signal and frequency-dependent ultrasonic group velocity and attenuation coefficient using wavelet analysis of ultrasonic echo waveform have been verified experimentally. The Gabor function is adopted the analyzing wavelet. The wavelet analysis shows that the variations of ultrasonic group velocity and attenuation coefficient due to the change of material characterization can be evaluated at each frequency. Furthermore, to assure the enhancement of detectability and naw sizing performance, both computer simulated results and experimental measurements using wavelet signal processing are used to demonstrate the effectiveness of the noise suppression of ultrasonic flaw signal obtained from austenitic stainless steel weld including EDM notch.

• The Journal of the Acoustical Society of Korea
• /
• v.25 no.7
• /
• pp.325-332
• /
• 2006

In this paper. we Propose the noise suppressor with the Perceptual model and wavelet packet transform. The objective is to enhance speech corrupted colored or non-stationary noise. If corrupted noise is colored. subband approach would be more efficient than whole band one. To avoid serious residual noise and speech distortion, we must adjust the Wavelet Coefficient Threshold (WCT). In this Paper. the subband is designed matching with the critical band and WCT is adapted noise masking threshold (NMT) and segmental signal to noise ratio (seg_SNR). Consequently. it has similar Performance with EVRC in PESQ-MOS. But it's better than wavelet packet transform using universal threshold about 0.289 in PESQ-MOS. The important thing is that it's more useful than EVRC in coded speech. In coded speech. PESQ-MOS is higher than EVRC about 0.23.

• Bulletin of the Korean Chemical Society
• /
• v.10 no.2
• /
• pp.123-128
• /
• 1989

Calorimetric study in conjunction with Fourier-transform infrared (FTIR) spectroscopic study was carried out on the blends of poly(ethylene oxide) (PEO) with isotactic, atactic and syndiotactic poly(methyl methacrylate) (i-, a-, and s-PMMA). From the differential scanning calorimetric (DSC) measurements, the three types of blends show a depression of the melting temperatures. This indicates that PEO is compatible with i-, a-, and s-PMMA. But the largest melting point depressions of PEO are always found in the blends with s-PMMA. For PEO/a-PMMA and PEO/s-PMMA, the degree of crystallinity as a function of composition deviates substantially from that of the ideal blend in which no interaction between the components exists. The FTIR spectra of all three types of blends are recorded. In order to observe the microstructural changes of PEO in blends, we analyzed the spectra using digital weighted subtraction and addition techniques. It was concluded that the microstructures of PEO are strongly perturbed by the PMMA's. Among these blends PEO microstructure in PEO/s-PMMA blends is most greatly influenced. It indicates that the blending is most preferred with s-PMMA than a- and i-PMMA. It can be explained on the basis of the molecular structure of PMMA's.

• Journal of Biosystems Engineering
• /
• v.34 no.2
• /
• pp.127-132
• /
• 2009

In the working population, muscle fatigue and musculoskeletal discomfort are common, which, in the case of insufficient recovery may lead to musculoskeletal pain. Workers suffering from musculoskeletal pains need to be rehabilitated for recovery. Isokinetic testing has been used in physical strengthening, rehabilitation and post-operative orthopedic surgery. Frequency analysis of electromyography (EMG) signals using the mean frequency (MNF) has been widely used to characterize muscle fatigue. During isokinetic contractions, EMG signals present strong nonstationarities. Hilbert-Haung transform (HHT) and autoregressive (AR) model have been known more suitable than Fourier or wavelet transform for nonstationary signals. Moreover, several analyses have been performed within each active phase during isokinetic contractions. Thus, the aims of this study were i) to determine which one was better suitable for the analysis of MNF between HHT and AR model during repetitive maximum isokinetic extensions and ii) to investigate whether the analysis could be repeated for sequential fixed epoch lengths. Seven healthy volunteers (five males and two females) performed isokinetic knee extensions at $60^{\circ}/s$ and $240^{\circ}/s$ until 50% of the maximum peak torque was reached. Surface EMG signals were recorded from the rectus femoris of the right thigh. An algorithm detecting the onset and offset of EMG signals was applied to extract each active phase of the muscle. Following the results, slopes from the least-square error linear regression of MNF values showed that muscle fatigue of all subjects occurred. The AR model is better suited than HHT for estimating MNF from nonstationary EMG signals during isokinetic knee extensions. Moreover, the linear regression can be extracted from MNF values calculated by sequential fixed epoch lengths (p> 0.0I).

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.19 no.11
• /
• pp.2094-2105
• /
• 1994

This paper contains a Hough transform based projection method derived from Radon transform for tracking dim unresolved(sub-pixel) moving targets that move along straight line parths across a time sequential image data. In contrast to several recently presented Hough transform methods using a compressed image referred to as the track map our proposed technique utilizing a set of projections taken along arbitrary orientations effectively increases the changes of target detection, and creates a robust track estimation environment by incorporating all the available knowledge obtained from the projections. Moreover, in order to quantitatively assess the estimation capability of the projection-based Hough transform algorithm, the analytical bounds on the Hough space parameter errors introduced by image space noise contamination are derived. The simulation yielded promising results of estimating the track parameters even under low signal to noise rations when our technique was tested against the time sequential sets of real infrared image data referred to as the HiCamps.