• The Journal of the Acoustical Society of Korea
• /
• v.14 no.3
• /
• pp.71-81
• /
• 1995

In this paper, feature vector extraction methods and classification algorithms for the automatic classification of transient signals in underwater are discussed. A feature vector extraction method using wavelet transform, which shows good performance with small number of coefficients, is proposed and compared with the existing classical methods. For the automatic classification, artificial neural networks such as multilayer perceptron (MLP), radial basis function (RBF), and MLP-Class are utilized, where those neural networks as well as extracted feature vectors are combined to improve the performance and reliability of the proposed algorithm. It is confirmed by computer simulation with Traco's standard transient data set I and simulated data that the proposed feature vector extraction method and classification algorithm perform well, assuming that the energy of a given transient signal is sufficiently larger than that of a ambient noise, that there are the finite number of noise sources, and that there does not exist noise sources more than two simultaneously.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.24 no.5
• /
• pp.566-573
• /
• 2013

Inverse synthetic aperture radar(ISAR) images have been widely used in non-cooperative target recognition(NCTR). One of the most important issues in ISAR imaging is the improvement of the image smeared by target motion. In this paper, we propose the discrete Gabor representation(DGR) in an oversampling scheme for efficient acquisition of high-quality ISAR images. The DGR compartmentally assigns the Gabor coefficients to unit cells of the time-frequency grid related to the given Gabor logons. Thus, it can show an excellent time-frequency concentration and effectively discriminates the Doppler components from point-scatterers. The simulation results demonstrated that the DGR not only obtained high-quality ISAR images but also retained computational efficiency.

• The Journal of the Korea Contents Association
• /
• v.4 no.1
• /
• pp.18-23
• /
• 2004

In this paper, we propose a new visual feature extraction method for content-based image retrieval(CBIR) based on wavelet transform which has both spatial-frequency characteristic and multi-resolution characteristic. We extract visual features for each frequency band in wavelet transformation and use them to CBIR. The lowest frequency band involves spacial information of original image. We extract L feature vectors using fuzzy homogeneity in the wavelet domain, which consider both the wavelet coefficients and the spacial information of each coefficient. Also, we extract 3 feature vectors wing the energy values of high frequency bands, and store those to image database. As a query, we retrieve the most similar image from image database according to the 10 largest homograms(normalized fuzzy homogeneity vectors) and 3 energy values. Simulation results show that the proposed method has good accuracy in image retrieval using 90 texture images.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.22 no.11
• /
• pp.2543-2551
• /
• 1997

This paper proposes a compression technique of image data, a variable length PN code and channel models which are required in CDMA communication system. It also analyzes their performances. Original images is compressed by 2-D DCT and its coefficients are quantized by optimal quantizer at compression rate 0.84bit/pel. Channel model 1 and 2 which are composed of 5 and 4 channels respectively are employed to be used in CDMA. Such a situation forces us to empoly variable length PN code, such as Chebyshev map for spread spectrum system. When average PN code length of model 1 and 2 is 44.4 and 26.7 chips respectively, the received image through these models under Gaussian noise with variance 1.75 is visually of the same quality as the transmitting image. Thus, the model 2 appears to be better in channel efficiency, comparing with channel model 1 and channel model which uses fixed length PN code.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.4
• /
• pp.31-37
• /
• 2004

Wavelet transform is a useful tool for analysis and process of image. This showed good performance in image compression and noise reduction. Wavelet coefficients can be effectively modeled by hidden Markov tree(HMT) model. However, in application of HMT model to image interpolation, training procedure is needed. Moreover, the parameters obtained from training procedure do not match input image well. In this paper, the structure of HMT is used for image interpolation, and the parameters of HMT are obtained from statistical characteristics across wavelet subbands without training procedure. In the proposed method, wavelet coefficient is modeled as Gaussian mixture model(GMM). In GMM, state transition probabilities are determined from statistical transition characteristic of coefficient across subbands, and the variance of each state is estimated using the property of exponential decay of wavelet coefficient. In simulation, the proposed method shows improvement of performance compared with conventional bicubic method and the method using HMT model with training.

• Fashion & Textile Research Journal
• /
• v.10 no.4
• /
• pp.566-571
• /
• 2008

Frictional sounds of 8 vapor permeable water repellent fabrics by sound generator were recorded and analyzed through FFT fast Fourier transform analysis. The frictional Sounds were quantified by calculating level pressure of total sound(LPT), the level range(${\Delta}L$) and the frequency difference(${\Delta}f$). Mechanical properties were measured by KES-FB. LPT values of specimens finished wet coating were higher than those of other kinds of finishing. ${\Delta}L$ values of specimens laminated were highest. Absolute values of ${\Delta}f$ were high in the cire finished and laminated specimens. Values for bending rigidity, shear stiffness and energy required for the compression of coated specimens increased compared with the cire finished and laminated specimens. Laminated specimens had high values of frictional coefficient and low values of surface roughness. Relationship between frictional sounds and mechanical properties analysed by use of correlation coefficients and stepwise regression. LPT showed significant correlation with elongation, tensile energy, geometrical roughness, weight and thickness. ${\Delta}L$ was highly correlated with tensile linearity, frictional coefficient, and ${\Delta}f$ with tensile linearity, weight and thickness. LPT were revealed to be explained by elongation and weight. ${\Delta}L$were predicted by tensile linearity, and ${\Delta}f$ by tensile linearity and thickness.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.48 no.2
• /
• pp.80-87
• /
• 2011

In this paper we present subject region-based auto-focusing algorithm using noise robust focus measure. The proposed algorithm automatically estimates the main subject using entropy and solves the traditional problems with a subject position or high frequency component of background image. We also propose a new focus measure by analyzing the discrete cosine transform coefficients. Experimental results show that the proposed method is more robust to Gaussian and impulse noises than the traditional methods. The proposed algorithm can be applied to Pan-tilt-zoom (PTZ) cameras in the intelligent video surveillance system.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.1
• /
• pp.76-84
• /
• 2007

The goal of this paper is to improve the watermarking performance using the following two methods; watermark estimation and low density parity check (LDPC) codes. For a blind watermark decoding, the power of a host image, which is hundreds times greater than the watermark power, is the main noise source. Therefore, a technique that can reduce the effect of the power of the host image to the detector is required. To this end, we need to estimate watermark from the watermarked image. In this paper, the watermark estimation is done by an adaptive estimation method with the generalized Gaussian distribution modeling of sub-band coefficients in the wavelet domain. Since the watermark capacity as well as the error rate can be improved by adopting optimum decoding principles and error correcting codes (ECC), we employ the LDPC codes for the decoding of the estimated watermark. Also, in LDPC codes, the knowledge about the noise power can improve the error correction capability. Simulation results demonstrate the superior performance of the proposed algorithm comparing to LDPC decoding with other estimation-based watermarking algorithms.

• Journal of the Korean Geotechnical Society
• /
• v.33 no.4
• /
• pp.17-24
• /
• 2017

Shear wave velocity (or shear modulus) of rock filled zone of CFRD is very important factor in the evaluation of performance of CFRD under the load such as earthquake. A shear wave velocity profile can be determined by surface wave method but this profile has been uncertainty caused by spatial variation of material property in rock filled zone. This uncertainty in shear wave velocity profile could be evaluated by the reliability based analysis which uses a coefficient of variation of material property to consider uncertainty caused by spatial variation of material property. In this paper, the possible 600 shear wave velocity profiles in rock filled zone of CFRD were generated using the method based on harmonic wavelet transform and 8 shear wave velocity profiles by HWAW method in the field, and the coefficients of variation of shear wave velocity with depth were evaluated for the rock filled zone of CFRD in Korea.

• The Journal of the Acoustical Society of Korea
• /
• v.15 no.2
• /
• pp.40-48
• /
• 1996

In this paper, a 40 dimensional segment vector with 4 frame and 7 frame width in every monosyllable interval was compressed into a 10, 14, 20 dimensional vector using K-L expansion and neural networks, and these was used to speech recognition feature parameter for CHMM. And we also compared them with CHMM added as feature parameter to the discrete duration time, the regression coefficients and the mixture distribution. In recognition test at 100 monosyllable, recognition rates of CHMM +${\bigtriangleup}$MCEP, CHMM +MIX and CHMM +DD respectively improve 1.4%, 2.36% and 2.78% over 85.19% of CHMM. And those using vector compressed by K-L expansion are less than MCEP + ${\bigtriangleup}$MCEP but those using K-L + MCEP, K-L + ${\bigtriangleup}$MCEP are almost same. Neural networks reflect more the speech dynamic variety than K-L expansion because they use the sigmoid function for the non-linear transform. Recognition rates using vector compressed by neural networks are higher than those using of K-L expansion and other methods.