• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.35-36
• /
• 2022

본 논문은 딥러닝 사람 자세 추정 모델이 사람의 관절 키포인트를 예측하는데 관절의 2차원 면적에 의해 키포인트별 𝜎, 즉, 표준 편차를 가지는 가우시안 커널(Gaussian Kernel)을 예측하는 방법을 제안한다. 각 관절 키포인트에 대해 다른 𝜎를 가지는 정답 히트맵(Ground Truth Heatmap)과 제안한 Gaussian Mixture Block를 모델에 추가해서 관절의 크기를 맞는 히트맵을 예측한다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.20 no.8
• /
• pp.1587-1595
• /
• 2016

P wave is cardiac parameters that represent the electrical and physiological characteristics, it is very important to diagnose atrial arrhythmia. However, It is very difficult to detect because of the small size compared to R wave and the various morphology. Several methods for detecting P wave has been proposed, such as frequency analysis and non-linear approach. However, in the case of conduction abnormality such as AV block or atrial arrhythmia, detection accuracy is at the lower level. We propose P wave detection algorithm through adaptive threshold and QRS peak variability. For this purpose, we detected Q, R, S wave from noise-free ECG signal through the preprocessing method. And then we classified three pattern of P wave by peak variability and detected adaptive window and threshold. The performance of P wave detection is evaluated by using 48 record of MIT-BIH arrhythmia database. The achieved scores indicate the average detection rate of 92.60%.

• Journal of KIISE:Computer Systems and Theory
• /
• v.35 no.1
• /
• pp.18-29
• /
• 2008

NAND Flash memory has been growing popular storage device for the last years because of its low power consumption, fast access speed, shock resistance and light weight properties. However, it has the distinct characteristics such as erase-before-write architecture, asymmetric read/write/erase speed, and the limitation on the number of erasure per block. Due to these limitations, various Flash Translation Layers (FTLs) have been proposed to effectively use NAND flash memory. The systems that adopted the conventional FTL may result in severe performance degradation by the hot data which are frequently requested data for overwrite in the same logical address. In this paper, we propose a novel FTL algorithm called Adaptive Flash Translation Layer (AFTL) which uses sector mapping method for hot data and log-based block mapping method for cold data. Our system removes the redundant write operations and the erase operations by the separating hot data from cold data. Moreover, the read performance is enhanced according to sector translation that tends to use a few read operations. A series of experiments was organized to inspect the performance of the proposed method, and they show very impressive results.

• The Journal of the Acoustical Society of Korea
• /
• v.7 no.4
• /
• pp.54-76
• /
• 1988

In Part Ⅰ of the paper, we have developed various block least mean-square (BLMS) adaptive digital filters (ADF's) based on a unified matrix treatment. In Part Ⅱ we analyze the convergence behaviors of the self-orthogonalizing frequency-domain BLMS (FBLMS) ADF and the unconstrained FBLMS (UFBLMS) ADF both for the overlap-save and overlap-add sectioning methods. We first show that, unlike the FBLMS ADF with a constant convergence factor, the convergence behavior of the self-orthogonalizing FBLMS ADF is governed by the same autocorrelation matrix as that of the UFBLMS ADF. We then show that the optimum solution of the UFBLMS ADF is the same as that of the constrained FBLMS ADF when the filter length is sufficiently long. The mean of the weight vector of the UFBLMS ADF is also shown to converge to the optimum Wiener weight vector under a proper condition. However, the steady-state mean-squared error(MSE) of the UFBLMS ADF turns out to be slightly worse than that of the constrained algorithm if the same convergence constant is used in both cases. On the other hand, when the filter length is not sufficiently long, while the constrained FBLMS ADF yields poor performance, the performance of the UFBLMS ADF can be improved to some extent by utilizing its extended filter-length capability. As for the self-orthogonalizing FBLMS ADF, we study how we can approximate the autocorrelation matrix by a diagonal matrix in the frequency domain. We also analyze the steady-state MSE's of the self-orthogonalizing FBLMS ADF's with and without the constant. Finally, we present various simulation results to verify our analytical results.

• The Journal of the Acoustical Society of Korea
• /
• v.7 no.4
• /
• pp.31-53
• /
• 1988

In this work we study extensively the structures and performance characteristics of the block least mean-square (BLMS) adaptive digital filters (ADF's) that can be realized efficiently using the fast Fourier transform (FFT). The weights of a BLMS ADF realized using the FFT can be adjusted either in the time domain or in the frequency domain, leading to the time-domain BLMS(TBLMS) algorithm or the frequency-domain BLMS (FBLMS) algorithm, respectively. In Part Ⅰof the paper, we first present new results on the overlap-add realization and the number-theoretic transform realization of the FBLMS ADF's. Then, we study how we can incorporate the concept of different frequency-weighting on the error signals and the self-orthogonalization of weight adjustment in the FBLMS ADF's , and also in the TBLMS ADF's. As a result, we show that the TBLMS ADF can also be made to have the same fast convergence speed as that of the self-orthogonalizing FBLMS ADF. Next, based on the properties of the sectioning operations in weight adjustment, we discuss unconstrained FBLMS algorithms that can reduce two FFT operations both for the overlap-save and overlap-add realizations. Finally, we investigate by computer simulation the effects of different parameter values and different algorithms on the convergence behaviors of the FBLMS and TBLMS ADF's. In Part Ⅱ of the paper, we will analyze the convergence characteristics of the TBLMS and FBLMS ADF's.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.46 no.3
• /
• pp.28-36
• /
• 2009

In this paper, we propose a frame-rate up-conversion method for temporal quality enhancement. The proposed method adaptively changes search range during hierarchical motion estimation and reconstructs hole regions using the proposed bi-direction prediction and linear interpolation. In order to alleviate errors due to inaccurate motion vector estimation, search range is adaptively changed based on reliability and for more accurate, motion estimation is performed in descending order of block variance. After segmentation of background and object regions, for filling hole regions, the pixel values of background regions are reconstructed using linear interpolation and those of object regions are compensated based on the proposed hi-directional prediction. The proposed algorithm is evaluated in terms of PSNR with original uncompressed sequences. Experimental results show that the proposed algorithm is better than conventional methods by around 2dB, and blocky artifacts and blur artifacts are significantly diminished.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.46 no.3
• /
• pp.89-95
• /
• 2009

H.264/AVC is the state-of-the-art video compression standard which achieves high coding efficiency compared with the previous standards. H.264/AVC adopts zig-zag scanning in order to encode quantized transform coefficients in a block. However, its performance is not satisfactory because all blocks are scanned in the fixed order without considering the characteristics of blocks. This paper presents an adaptive coefficient scanning method for improving inter coding efficiency in H.264/AVC. In the proposed method, the coefficient scanning order for each prediction mode is adaptively controlled based on the information of previously-coded blocks. The experimental results show that the proposed coefficient scanning method improves the coding efficiency about 2.29% for high-quality HD sequences.

• Journal of IKEEE
• /
• v.13 no.1
• /
• pp.1-10
• /
• 2009

Recently, some frequency domain CMA (Constant Modulus Algorithm) have been introduced in an effort to reduce computational complexities [1,2]. In [1], a fast algorithm that minimizing cost function designed for block input signal is employed, while in [2], a novel cost function that minimizing sample by sample input is used. Although, the two fast algorithm save computational complexities as compared to CMA, the convergence behaviors of the two fast algorithm show different results with repsect to CMA. Thus, in this paper, some analytical results on the error surface of the fast frequency domain CMA are introduced. From the analytical results, we show that the more recent algorithm [2] outperforms the previous algorithm [1]. Simulation results reveals that the recent algorithm [2] shows 50% enhanced convergence with respect to the old fast algorithm [1]. Also, we show that the recent fast algorithm [2] has comparable convergence performance with respect to conventional CMA algorithm.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38B no.2
• /
• pp.154-161
• /
• 2013

This paper presents the signal-to-interference plus noise ratio (SINR) estimation of IEEE 802.16m WirelessMAN-Advanced mobile station with simulation and implementation results. The downlink signal of IEEE 802.16m has two kinds of A-Preambles: the PA-preamble and the SA-preamble. This paper proposes the efficient method of estimating SINR with A-Preambles, by measuring noise power from PA-preamble and measuring interference power and signal power from SA-preamble. The proposed SINR measurement block contains important features such as subcarrier phase rotation elimination and simplified dB transform. The result of this paper is integrated to ETRI's IEEE 802.16m test mobile station, used for decision of adaptive-modulation-and-coding (AMC) and hand-over. It showed good measurement performance in simulation and unified system link test also.

• Transactions on Control, Automation and Systems Engineering
• /
• v.4 no.3
• /
• pp.224-230
• /
• 2002

In this paper, a counting algorithm hybridized with an adaptive automatic thresholding method based on Otsu's method and the algorithm that elongates markers obtained by the well-known watershed algorithm is proposed to enhance the exactness of the microcell counting in microscopic images. The proposed counting algorithm can be stated as follows. The transformed full image captured by CCD camera set up at microscope is divided into cropped images of m$\times$n blocks with an appropriate size. The thresholding value of the cropped image is obtained by Otsu's method and the image is transformed into binary image. The microbial cell images below prespecified pixels are regarded as noise and are removed in tile binary image. The smoothing procedure is done by the area opening and the morphological filter. Watershed algorithm and the elongating marker algorithm are applied. By repeating the above stated procedure for m$\times$n blocks, the m$\times$n segmented images are obtained. A superposed image with the size of 640$\times$480 pixels as same as original image is obtained from the m$\times$n segmented block images. By labeling the superposed image, the counting result on the image of microbial cells is achieved. To prove the effectiveness of the proposed mettled in counting the microbial cell on the image, we used Acinetobacter sp., a kind of ammonia-oxidizing bacteria, and compared the proposed method with the global Otsu's method the traditional watershed algorithm based on global thresholding value and human visual method. The result counted by the proposed method shows more approximated result to the human visual counting method than the result counted by any other method.