• ETRI Journal
• /
• 제45권6호
• /
• pp.1035-1045
• /
• 2023

We propose a novel graphics processing unit (GPU) algorithm that can handle a large-scale 3D fast Fourier transform (i.e., 3D-FFT) problem whose data size is larger than the GPU's memory. A 1D FFT-based 3D-FFT computational approach is used to solve the limited device memory issue. Moreover, to reduce the communication overhead between the CPU and GPU, we propose a 3D data-transposition method that converts the target 1D vector into a contiguous memory layout and improves data transfer efficiency. The transposed data are communicated between the host and device memories efficiently through the pinned buffer and multiple streams. We apply our method to various large-scale benchmarks and compare its performance with the state-of-the-art multicore CPU FFT library (i.e., fastest Fourier transform in the West [FFTW]) and a prior GPU-based 3D-FFT algorithm. Our method achieves a higher performance (up to 2.89 times) than FFTW; it yields more performance gaps as the data size increases. The performance of the prior GPU algorithm decreases considerably in massive-scale problems, whereas our method's performance is stable.

• Journal of the Korean Data and Information Science Society
• /
• 제10권1호
• /
• pp.243-250
• /
• 1999

We consider $Y=X{\lambda}Z,\;{\lambda}>0$, where X and Z are independent random variables, and Y is the length biased distribution or the equilibrium distribution of X. The purpose of this paper is to consider the distribution of X or Y when the distribution of Z is given and the distribution of Z when the distribution of X or Y is given, In particular, we obtain that the necessary and sufficient conditions for X to be $X^{2}({\upsilon})\;is\;Z{\sim}X^{2}(2)\;and\;for\;Z\;to\;be\;X^{2}(1)\;is\;X{\sim}IG({\mu},\;{\mu}^{2}/{\lambda})$, where $IG({\mu},\;{\mu}^{2}/{\lambda})$ is two-parameter inverse Gaussian distribution. Also we show that X is smaller than Y in the reverse Laplace transform ratio order if and only if $X_{e}$ is smaller than $Y_{e}$ in the Laplace transform ratio order. Finally, we can get the results that if X is smaller than Y in the Laplace transform ratio order, then $Y_{L}$ is smaller than $X_{L}$ in the Laplace transform order, and that if X is smaller than Y in the reverse Laplace transform ratio order, then $_{\mu}X_{L}$ is smaller than $_{\nu}Y_{L}$ in the Laplace transform order.

• Smart Structures and Systems
• /
• 제1권2호
• /
• pp.185-215
• /
• 2005

Advanced signal processing techniques have been long introduced and widely used in structural health monitoring (SHM) and nondestructive evaluation (NDE). In our research, we applied several signal processing approaches for our embedded ultrasonic structural radar (EUSR) system to obtain improved damage detection results. The EUSR algorithm was developed to detect defects within a large area of a thin-plate specimen using a piezoelectric wafer active sensor (PWAS) array. In the EUSR, the discrete wavelet transform (DWT) was first applied for signal de-noising. Secondly, after constructing the EUSR data, the short-time Fourier transform (STFT) and continuous wavelet transform (CWT) were used for the time-frequency analysis. Then the results were compared thereafter. We eventually chose continuous wavelet transform to filter out from the original signal the component with the excitation signal's frequency. Third, cross correlation method and Hilbert transform were applied to A-scan signals to extract the time of flight (TOF) of the wave packets from the crack. Finally, the Hilbert transform was again applied to the EUSR data to extract the envelopes for final inspection result visualization. The EUSR system was implemented in LabVIEW. Several laboratory experiments have been conducted and have verified that, with the advanced signal processing approaches, the EUSR has enhanced damage detection ability.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1996년도 추계학술대회 논문집 학회본부
• /
• pp.39-41
• /
• 1996

While the z-transform method is a basic mathematical tool to relate the signals only at the sampling instants in analyzing and designing sampled-data control systems, the modified z-transform which is a variation of the z-transform is widely used to represent the details of continuous signals between the sampling instants. Regarding the modified z-transform method, some properties were established to relate the modified z-transform to the regular z-transform. This paper will show that these properties, in their current forms, cause some analytic problems, when they are applied to the signals with discontinuities at the sampling instants, which accordingly limit their applications significantly. In this paper, those analytic problems will be investigated, and the theorems of the modified z-transform will be revised by adopting a new notation so that those can be correctly interpreted and used without any analytic problems in the analysis of sampled data systems. Also some useful schemes of applying the modified z-transform will be developed.

• 한국통신학회논문지
• /
• 제11권3호
• /
• pp.197-203
• /
• 1986

아날로그 映像信號를 디지털 變換하여 마이크로컴퓨터에 入力시킨 다음 變換符號化(trarm coding)를 遂行하였다. 入力 및 處理 時에 遂行時間을 短縮시키는 것을 目的으로 하여 64$\times$64화소로 信號를 滅縮하고 外部메모리를 內部메모리의 番地로 對替시키는 것으로써 빠른 入出力 方法을 채택하였다. Hadamard 直交 變換과 中央에 weight를 준 weighted hadamard變換 그리고 haar 變換을 assembly言語의 高速計算 알고리즘으로 프로그램하여 一秒 이내에 變換終了할 수 있도록 하였으며 各 變換을 比較하였다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 1999년도 추계종합학술대회 논문집
• /
• pp.609-612
• /
• 1999

In this paper, we propose wavelet transform image compression method such that an image is decomposed into multiresolutions using biorthogonal wavelet transform with linear phase response property and decomposed subbands are classified by maximum classification gain. The classified data is quantized by allocating bits in accordance with classified class informations within subbands through arbitrary set bit allocation algorithm. And then, quantized data in each subband are entropy coded. The proposed coding method is that the quantized data perform shuffling before entropy coding in order to remove sign bit plane. And the context is assigned by maximum correlation direction for bit plane coding.

• 한국원자력학회:학술대회논문집
• /
• 한국원자력학회 1997년도 추계학술발표회논문집(1)
• /
• pp.225-230
• /
• 1997

Discrete Wavelet Transforms (DWTs) are recent mathematics, and begin to be used in various fields. The wavelet transform can be used to compress the signal and image due to its inherent properties. We applied the wavelet transform compression and reconstruction to the neutron cross section data. Numerical tests illustrate that tile signal compression using wavelet is very effective to reduce the data saving spaces.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2002년도 하계종합학술대회 논문집(4)
• /
• pp.227-230
• /
• 2002

In this paper, we propose an algorithm to decode Data Matrix two dimensional barcode symbology. We employ hough transform and bilinear image warping to extract the barcode region from the image scanned using a CMOS digital camera. The location of barcode can be found by applying Hough transform. However, barcode image should be warped due to the nonlinearity of lens and the viewing angle of camera. In this paper, bilinear warping transform is adopted to wa게 and align the barcode region of the scanned image. Codeword can be detected from the aligned barcode region.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 1995년도 추계학술대회 논문집
• /
• pp.361-364
• /
• 1995

Wavelet Transform is a new tools for signal processing, such as data compressing extraction of parameter for Reconition and Diagnostics. This transform has an advandage of a good resolution compared to Fast Fourier Transform (FFT) In this study, we employ the wavelet transform for analysis of Acoustic Emission raw signal generated form rotary compressor. In abnormal condition of rotary compressor, the state of operating condition can be classified by analizing coefficient of wavelet transformed signal.

• 한국통신학회논문지
• /
• 제11권4호
• /
• pp.251-258
• /
• 1986

TV와 같은 영상신호에는 중복도가 많이 존재하며, 이를 줄이기 위한 여러가지 방법들이 연구되고 있다. 본 논문에서는 Hadamard 변환을 이용하여 computer simulation과 실험 모델을 제작하여 데이터 압축에 관해 연구하였다. Hadamard matrix는 +1과 -1로 구성되며, row vector들은 서로 orthogonal하고 변환된 signal을 계산하기 위해서는 가산과 감산만이 필요하기 때문에 가산뿐 아니라 승산이 필요한 Fourier transform등 다른 orthogonal transform 에 비해 hardware구성이 용이하다. 링컨데이타 (64$ imes$64)를 8차와 16차 Hadamard 변환으로 simulation하였고, 8차를 hardware로 구성하였으며 이 경우 이론과 실험을 통해 연구한 결과 좋은 화질을 얻기 위해서는 2.0bits/sample가 필요했다.