• Journal of the Institute of Electronics Engineers of Korea TE
• /
• v.39 no.3
• /
• pp.83-89
• /
• 2002

A new scheme for image compression based on texture expression in the wavelet transform domain is presented. After taking wavelet transform, using the fact that the high-pass filtered bands has a lower variance than that of the original, a texture expression for the homogeneous polygonal regions can be more efficiently performed in the wavelet transform domain. The estimated texture parameters are transmitted to the receiver and later used for reconstruction after storing in disk. In most cases, the proposed method has yields good results with respects to the compression ratio and reconstructed image quality when our system has compared to conventional SPIHT scheme. 

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.29 no.3C
• /
• pp.387-397
• /
• 2004

The wavelet transform providing both of the frequency and spatial information of an image is proved to be very much effective for the compression of images, and recently lot of studies on coding algorithms for images decomposed by the wavelet transform together with the multi-resolution theory are going on. This paper proposes a quadtree decomposition method of image compression applied to the images decomposed by wavelet transform by using the correlations between pixels and '0'data grouping. Since the coefficients obtained by the wavelet transform have high correlations between scales and high concentrations, the quadtree method can reduce the data quantity effectively. the experimental infrared image with 256${\times}$256 size and 8〔bit〕, was used to compare the performances of the existing and the proposed compression methods.

• Proceedings of the IEEK Conference
• /
• 1998.06a
• /
• pp.745-748
• /
• 1998

Recently, image and voice processing part is using wavelet transform. We propose thining algorithm using wavelet tranform. Wavelet transform consists of low frequency and high frequency in the spatial and frequency domain. After the wavelet decomposition, more than 90 percents of energy are contained in lowest frequency band. Therefor, for images with large difference of gray value between foreground and background like character images, we can more accurately in the lowest frequency band. Lowest frequency band has wavelet transform significant coefficient(WTS) that is required for the thinning algorithm we proposed Paper [3][5][7][8] can not separate consonants and vowels of korean characters. Becuase korean characters have structural feature. This paper can separate consonants and vowels. Simulation executed low frequency image and data compression can reduce 1/4$^{n}$ with level n. we can redcue time complexity 3/8.

• The Transactions of the Korea Information Processing Society
• /
• v.5 no.4
• /
• pp.1048-1056
• /
• 1998

In this paper, a fractal image coding based on discrete wavelet transform is proposed to improve PSNR at low bit rates and reduce computational complexity of encoding process. The proposed method takes the absolute value of discrete wavelet transform coefficients, and then constructs significant coefficients trees, which indicate the positions and signs of the significant coefficients. This method improves PSNR and reduces computational complexity of mapping contracted domain pool onto range block, by matching only the significant coefficients of range block to coefficients of contracted domain block. Also, this paper proposes a classification scheme which minimizes the number of contracted domain blocks compared with range block. This scheme significantly reduces the number of range and contracted domain block comparison.

• Journal of Korea Multimedia Society
• /
• v.4 no.4
• /
• pp.306-315
• /
• 2001

In noise reduction method from noisy speech for speech recognition in noisy environments, conventional spectral subtraction method has a disadvantage which distinction of noise and speech is difficult, and characteristic of noise can't be estimated accurately. Also, noise reduction method in the wavelet transform domain has a disadvantage which loss of signal is generated in the high frequency domain. In order to compensate theme disadvantage, this paper propose spectral subtraction method in continuous wavelet transform domain which speech and non- speech intervals is distinguished by standard deviation of wavelet coefficient, and signal is divided three scales at different scale. The proposed method extract accurately characteristic of noise in order to apply spectral subtraction method by end detection and band division. The proposed method shows better performance than noise reduction method using conventional spectral subtraction and wavelet transform from viewpoint signal to noise ratio and Itakura-Saito distance by experimental.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.18 no.6
• /
• pp.1103-1108
• /
• 2023

In this paper, we propose a new algorithm for attenuating the background noises in acoustic signal. This algorithm improves the noise attenuation performance by using the FNN(: Full-connected Neural Network) deep learning algorithm instead of the existing adaptive filter after wavelet transform. After wavelet transforming the input signal for each short-time period, noise is removed from a single input audio signal containing noise by using a 1024-1024-512-neuron FNN deep learning model. This transforms the time-domain voice signal into the time-frequency domain so that the noise characteristics are well expressed, and effectively predicts voice in a noisy environment through supervised learning using the conversion parameter of the pure voice signal for the conversion parameter. In order to verify the performance of the noise reduction system proposed in this study, a simulation program using Tensorflow and Keras libraries was written and a simulation was performed. As a result of the experiment, the proposed deep learning algorithm improved Mean Square Error (MSE) by 30% compared to the case of using the existing adaptive filter and by 20% compared to the case of using the STFT(: Short-Time Fourier Transform) transform effect was obtained.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.4
• /
• pp.1121-1131
• /
• 1997

In video coding at high compression rate, factal compression schemes in spatial domain have outstanding blocking artifacts and compression schemes in wavelet transform domain have rinfing artifacts at edges. In order to compensate these disadvantages, we propose a fractal video coding in wavelet transrorm domain which leads to clear edges without blocking atrifacts even at high bompression rate. The proposed method performs variable block sized motion estimation by using correlation among different subbands. Then the wavelet coefficients which are not enoded dffectively by the motion estimation are compressed by inter-frame fractal coding which predicts fine scale subbands hierarchically from the next coarser scale subbands. Computer sumulations with sev-eral test images wequences show that the proposed method shows better performance than the conventional video coding methods using fractal and wavelet.

• Proceedings of the Korea Contents Association Conference
• /
• 2005.11a
• /
• pp.181-185
• /
• 2005

In this paper, the DWT(Discrete Wavelet Transform) based watermarking method for authentication and detection of image modification was proposed. The proposed algorithm inserts watermark into high frequency domain after 1-level wavelet transform by exchanging wavelet coefficients and embeds the characteristic values of high frequency domain of original image into the LSB part of watermarked image. Therefore, By extracting LSB values and watermark in the high frequency domain from the watermarked image, we can authenticate the image and detect modified positions.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.18 no.10
• /
• pp.1469-1482
• /
• 1993

A new image coding technique has been proposed based on the wavelet transform. To achieve lower ceding rates and good qualities in reconstructed images, some of wavelet coefficients were removed by thresholding and quantized in accordance with the sensitivity of the human visual system(HVS). For each block of subimages in wavelet transform domain, block thresholding scheme has been used to remove the unimportant wavelet coefficients according to the frequency characteristic and statistical property of wavelet coefficients. The location information of quantized blocks and removed blocks were encoded using run-length coder which is effective for the exponential distribution. Quantized coefficients were encoded using variable length coder which matches well to their distribution. Simulation results show that the reconstructed images maintain high quality with the low bit rate, below 1.0 bits per pel.

• Journal of the Korean Institute of Telematics and Electronics T
• /
• v.35T no.3
• /
• pp.37-45
• /
• 1998

In this paper, the different frequency component and time informations from an EMI signal are extracted simultaneously using a wavelet transform and the results of transform in the time and frequency domain are analyzed. Frequencies are extracted from the EMI signal by performing the multiresolution analysis using the Daubechies-4 filter coefficients and the time information through the results of wavelet transform. We have tried the correlation analysis to evaluate the results of wavelet transform. We have chosen the optimal wavelet function for an object signal by comparing the transformed results of various wavelet functions and verified the simulation examples of waveform and harmonic analysis using a wavelet transform. We have proved the denoising effect to the EMI signal using the soft thresholding technique.