• Journal of KSNVE
• /
• v.11 no.3
• /
• pp.455-460
• /
• 2001

An exact spectrum wish no leakage error could be obtained when the period of the signal coincides perfectly with the record length. However, the record length will be determined regardless of the period of signal. The Leakage error due to this problem will gibe a distorted spectrum. In the conventional research, the method was proposed to estimate the three parameters, frequency, amplitude and phase angle, from the spectrum data for anundamped sinusoidal signal. In this paper, some techniques are proposed to estimate frequency, amplitude and damping ratios from the frequency response functions for damped signals. The validation of the proposed techniques is verified by several numerical examples.

• Biomolecules & Therapeutics
• /
• v.8 no.1
• /
• pp.63-72
• /
• 2000

We intended to show some basic requirements for spectrum analysis of electroencephalogram (EEG) by visualizing the differences of the results according to different values of some parameters for analysis. Spectrum analysis is the most popular technique applied for the quantitative analysis of the electroen- cephalographic signals. Each step from signal acquisition through spectrum analysis to presentation of parameters was examined with providing some different values of parameters. The steps are:(1) signal acquisition; (2) spectrum analysis; (3) parameter extractions; and (4) presentation of results. In the step of signal acquisition, filtering and amplification of signal should be considered and sampling rate for analog-to-digital conversion is two-time faster than highest frequency component of signal. For the spectrum analysis, the length of signal or epoch size transformed to a function on frequency domain by courier transform is important. Win dowing method applied for the pre-processing before the analysis should be considered for reducing leakage problem. In the step of parameter extraction, data reduction has to be considered so that statistical comparison can be used in appropriate number of parameters. Generally, the log of power of all bands is derived from the spectrum. For good visualization and quantitative evaluation of time course of the parameters are presented in chronospectrogram.

• Korean Journal of Remote Sensing
• /
• v.14 no.1
• /
• pp.17-36
• /
• 1998

The Radon transform has been widely used in various techniques of digital image processing such as the computerized topography, lineament analysis in a remotely sensed image, slant-stack processing of seismic data, and so on. Compared to the Fourier transform, the utility of two-dimensional convolutional or correlational properties of the Radon transform, however, has been underestimated. We show that the two-dimensional convolution and correlation is respectively reduced to be one-dimensional convolution and correlation with respect to ρ in the Radon space. Therefore, one can achieve a two dimensional filtering by applying a simple one-dimensional convolution in the Radon space followed by an inverse Radon transform. Tests of the approach using FIR filters are carried out specifically for enhancing the ship wake in a RADARSAT SAR image. The test results demonstrate that the two-dimensional filtering through the Radon transform effectively enhance the ship wake features as well as reducing sea speckle in the image. Although two-dimensional convolution and correlation through the Radon transform are not so much useful as those through the courier transform in views of efficiency and effectiveness, it can be utilized to improve the quality of a digitally processed output when the process should be accompanied by the Radon transform such as topography and lineament analysis of SAR image.