• Journal of Power Electronics
• /
• v.17 no.2
• /
• pp.358-367
• /
• 2017

This paper presents a novel filtering method for speed measurements to improve the low-speed performance of the direct-drive permanent magnet traction machines for elevators. Based on the theory of nonlinear tracking differentiator (NTD), this method, which can act as a high performance filter of a raw speed signal, obtains a more accurate speed feedback signal when applying a low-resolution encoder. In addition, it can relieve the interference caused by the position derivative for speed sampling. By analyzing the frequency response of the NTD, the influence of its parameters on the performance of the speed filtering is investigated. Compared with different types of low-pass filters, the proposed method shows a shorter time delay and a stronger ability in terms of noise suppression when the parameters are selected carefully. In addition, when using the measured speed signal through a nonlinear tracking differentiator as the feedback of the system, the motor runs more steadily at low speeds. As a result, the riding comfort of a direct-drive elevator can be improved. The feasibility of the proposed strategy was verified on an 11.7kW elevator traction machine using a commercial inverter.

• Journal of KIISE:Software and Applications
• /
• v.31 no.9
• /
• pp.1150-1160
• /
• 2004

In this paper, we propose a new 3D video coding method for heterogeneous display systems and network infrastructure over enhanced Access Grid (e-AG) using spatio-temporal scalability defined in MPEG-2. The proposed encoder produces several bit-streams for providing temporally and spatially scalable 3D video service. The generated bit-streams can be nelivered with proper spatio-temporal resolution according to network bandwidths and processing speeds, visualization capabilities of client systems. The functionality of proposed spatio-temporal scalability can be exploited for construction of highly scalable 3D video service in heterogeneous distributed environments.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.7C
• /
• pp.700-705
• /
• 2006

In this paper, an informed watermarking algorithm is proposed that aids in concealing packet loss errors in video communications. This watermark-based error concealment (WEC) method embeds a low resolution version of the video frame inside itself as watermark data. At the receiver, the extracted watermark is used as a reference for error concealment. The proposed DCT-based algorithm employs informed watermarking techniques in order to minimize the distortion of host frames. At the encoder, a predictive feedback loop is employed which helps to adjust the strength of the data embedding. Furthermore, the distortion of the DCT coefficients introduced in the embedding can be removed to a considerable extent, by employing bit-sign adaptivity. Simulation results on standard video sequences show that the proposed informed WEC scheme has an advantage of 3$\sim$4 dB in PSNR over non-informed WEC and that even a non-informed WEC is still superior to conventional error concealment techniques.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.10 no.5
• /
• pp.942-946
• /
• 2009

A scan order converter has to be placed before the JPEG encoder to provide $8{\times}8$ blocks from the pixels in raster scan order. Recently a hardware architecture has been proposed to implement a scan converter based on the single line memory. Since both read and write accesses happen at each cycle, however, the largest part of the entire power budget is occupied by the SRAM itself. In this paper, the data packing and unpacking procedure is inserted in the processing chain, such that the access frequency to the SRAM is reduced to 1/8 by adopting a packed larger data unit. The simulation results show that the resultant power consumption is reduced down to 16% for the SXGA resolution.

• Proceedings of the KIEE Conference
• /
• 1995.07b
• /
• pp.846-848
• /
• 1995

Recently, direct drive AC servo motor and DSP is widely used in the factory automation and related field due to the enhanced control and digital technology. In this paper, position control of direct drive AC servo motor and numerical interpolation algorithms are studied using DSP for the future applications. In the direct drived motor control, encoder pulse is divided into the quadruple pulse for the higher resolution and high speed pulse(10MHz) is used as reference for the slower speed motor control (M/T method). As for the general position control, PI controller is adapted for position control. In the trajectory tracking numerical algorithm, interpolation of straight line and curve algorithms are studied for the realization of path following capabilities of XY table. As for the DSP, ADSP 2105 is selected for the economy and performance points of view.

• ETRI Journal
• /
• v.41 no.3
• /
• pp.371-382
• /
• 2019

Machine reading comprehension is the task of understanding a given context and finding the correct response in that context. A simple recurrent unit (SRU) is a model that solves the vanishing gradient problem in a recurrent neural network (RNN) using a neural gate, such as a gated recurrent unit (GRU) and long short-term memory (LSTM); moreover, it removes the previous hidden state from the input gate to improve the speed compared to GRU and LSTM. A self-matching network, used in R-Net, can have a similar effect to coreference resolution because the self-matching network can obtain context information of a similar meaning by calculating the attention weight for its own RNN sequence. In this paper, we construct a dataset for Korean machine reading comprehension and propose an $S^2-Net$ model that adds a self-matching layer to an encoder RNN using multilayer SRU. The experimental results show that the proposed $S^2-Net$ model has performance of single 68.82% EM and 81.25% F1, and ensemble 70.81% EM, 82.48% F1 in the Korean machine reading comprehension test dataset, and has single 71.30% EM and 80.37% F1 and ensemble 73.29% EM and 81.54% F1 performance in the SQuAD dev dataset.

• Korean Journal of Artificial Intelligence
• /
• v.11 no.3
• /
• pp.17-22
• /
• 2023

In this paper, we propose a black ice detection system using Convolutional Neural Networks (CNNs). Black ice poses a serious threat to road safety, particularly during winter conditions. To overcome this problem, we introduce a CNN-based architecture for real-time black ice detection with an encoder-decoder network, specifically designed for real-time black ice detection using thermal images. To train the network, we establish a specialized experimental platform to capture thermal images of various black ice formations on diverse road surfaces, including cement and asphalt. This enables us to curate a comprehensive dataset of thermal road black ice images for a training and evaluation purpose. Additionally, in order to enhance the accuracy of black ice detection, we propose a multi-scale dilation convolution feature fusion (MsDC-FF) technique. This proposed technique dynamically adjusts the dilation ratios based on the input image's resolution, improving the network's ability to capture fine-grained details. Experimental results demonstrate the superior performance of our proposed network model compared to conventional image segmentation models. Our model achieved an mIoU of 95.93%, while LinkNet achieved an mIoU of 95.39%. Therefore, it is concluded that the proposed model in this paper could offer a promising solution for real-time black ice detection, thereby enhancing road safety during winter conditions.

• Journal of Broadcast Engineering
• /
• v.22 no.1
• /
• pp.28-41
• /
• 2017

In this paper, the fast encoding algorithm in High Efficiency Video Coding (HEVC) encoder was studied. For the encoding efficiency, the current HEVC reference software is divided the input image into Coding Tree Unit (CTU). then, it should be re-divided into CU up to maximum depth in form of quad-tree for RDO (Rate-Distortion Optimization) in encoding precess. But, it is one of the reason why complexity is high in the encoding precess. In this paper, to reduce the high complexity in the encoding process, it proposed the method by determining the maximum depth of the CU using a hierarchical clustering at the pre-processing. The hierarchical clustering results represented an average combination of motion vectors (MV) on neighboring blocks. Experimental results showed that the proposed method could achieve an average of 16% time saving with minimal BD-rate loss at 1080p video resolution. When combined the previous fast algorithm, the proposed method could achieve an average 45.13% time saving with 1.84% BD-rate loss.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.43 no.5 s.347
• /
• pp.1-10
• /
• 2006

In this paper, an 1.8V 8-bit 500MSPS CMOS A/D Converter is proposed. In order to obtain the resolution of 8bits and high-speed operation, a Cascaded-Folding Cascaded-Interpolation type architecture is chosen. For the purpose of improving SNR, Cascaded-folding Cascaded-interpolation technique, distributed track and hold are included [1]. A novel folding circuit, a novel Digital Encoder, a circuit to reduce the Reference Fluctuation are proposed. The chip has been fabricated with a $0.18{\mu}m$ 1-poly 5-metal n-well CMOS technology. The effective chip area is $1050{\mu}m{\times}820{\mu}m$ and it dissipates about 146mW at 1.8V power supply. The INL and DNL are within ${\pm}1LSB$, respectively. The SNDR is about 43.72dB at 500MHz sampling frequency.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.12
• /
• pp.150-159
• /
• 2014

The latest video coding standard, HEVC can improve the coding efficiency significantly compared with the H.264/AVC. However the HEVC encoder requires much larger computational complexities. The longer 8-tap interpolation filter of the HEVC which is used in a non-integer motion estimation is one of the reasons and this paper aims to reduce the computational complexities. First of all, three shorter-tap interpolation filters for a motion estimation process are tested rather than the use of a standard interpolation filter. In addition, the fast searching strategies to reduce the number of comparisons for choosing the best non-integer motion vector are proposed. Finally, the interpolation process is selectively applied according to the searching strategy. By combining all of the techniques, the experimental results show that the encoding times can be reduced by 13.6%, 18.5% and 21.1% with the coding efficiency penalties of 0.7%, 1.5% and 2.5%, respectively. For the full-HD video sequences, the coding efficiency penalties are reduced to 0.4%, 1.1% and 1.6% at the same level of the encoding time savings, which shows the effectiveness of the proposed schemes for the high resolution video sequences.