• Journal of Korea Game Society
• /
• v.14 no.6
• /
• pp.59-68
• /
• 2014

Memory bandwidth requirements of UHD (Ultra High Definition $4096{\times}2160$) game scenes have been much more increasing. This paper presents a lossless DPCM-GR based compression algorithm using CUDA for solving the memory bandwidth problem without sacrificing image quality, which is modified from DDPCM-GR [4] to support bit parallel pipelining. The memory bandwidth efficiency increases because of using the shared memory of CUDA. Various asynchronous transfer configurations which can overlap the kernel execution and data transfer between host and CUDA are implemented with the page-locked host memory. Experimental results show that the maximum 31.3 speedup is obtained according to CPU time. The maximum 30.3% decreases in the computation time among various configurations.

• Journal of Internet Computing and Services
• /
• v.18 no.6
• /
• pp.1-6
• /
• 2017

For the algorithm of cross-information potential with delta functions (CIPD) which has superior performance in impulsive noise environments, a new method of employing the information of power variations of magnitude controlled input (MCI) in the weight update equation of the CIPD is proposed in this paper where the input of CIPD is modified by the Gaussian kernel of error. To prove its effectiveness compared to the conventionalCIPD algorithm, the distance between the current weight vector and its previous one is analyzed and compared under impulsive noise. In the simulation results the proposed method shows a two-fold improvement in steady state stability, faster convergence speed by 1.8 times, and 2 dB - lower minimum MSE in the impulsive noise situation.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.60 no.1
• /
• pp.43-49
• /
• 2011

A distance relay scheme is commonly used for backup protection. This scheme, called a step distance protection, is comprised of 3 steps for graded zones having different operating time. As for the conventional step distance protection scheme, Zone 2 can exceed the ordinary coverage excessively in case of a transformer protection relay especially. In this case, there can be overlapped protection area from a backup protection relay and, therefore, malfunctions can occur when any fault occurs in the overlapped protection area. Distance relays and overcurrent relays are used for backup protection generally, and both relays have normally this problem, the maloperation, caused by a fault in the overlapped protection area. Corresponding to an IEEE standard, this problem can be solved with the modification of the operating time. On the other hand, in Korea, zones are modified to cope with this problem in some specific conditions. These two methods may not be obvious to handle this problem correctly because these methods, modifying the common rules, can cause another coordination problem. To overcome this problem clearly, this paper describes an improved backup protection coordination scheme using an IEC 61850-based distance relay for transformer backup protection. IEC 61850-based IED(Intelligent Electronic Device) and the network system based on the kernel 2.6 LINUX are realized to verify the proposed method. And laboratory tests to estimate the communication time show that the proposed coordination method is reliable enough for the improved backup protection scheme.

• Transactions of the Korean Society of Pressure Vessels and Piping
• /
• v.19 no.2
• /
• pp.146-154
• /
• 2023

In nuclear power plants, ultrasonic test for pipe wall thickness measurement is used during periodic inspections to prevent pipe rupture due to pipe wall thinning. However, when measuring pipe wall thickness using ultrasonic test, a significant amount of measurement error occurs due to the on-site conditions of the nuclear power plant. If the maximum pipe wall thinning rate is decided by the measured pipe wall thickness containing a significant error, the pipe wall thinning rate data have significant uncertainty and systematic overestimation. This study proposes preprocessing of pipe wall thinning measurement data using support vector machine regression algorithm. By using support vector machine, pipe wall thinning measurement data can be smoothened and accordingly uncertainty and systematic overestimation of the estimated pipe wall thinning rate data can be reduced.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.48 no.6
• /
• pp.8-17
• /
• 2011

Recently, fusion camera systems that consist of depth sensors and color cameras have been widely developed with the advent of a new type of sensor, time-of-flight (TOF) depth sensor. The physical limitation of depth sensors usually generates low resolution images compared to corresponding color images. Therefore, the pre-processing module, such as camera calibration, three dimensional warping, and hole filling, is necessary to generate the high resolution depth map that is placed in the image plane of the color image. However, the result of the pre-processing step is usually inaccurate due to errors from the camera calibration and the depth measurement. Therefore, in this paper, we present a depth map upsampling method robust these errors. First, the confidence of the measured depth value is estimated by the interrelation between the color image and the pre-upsampled depth map. Then, the detailed depth map can be generated by the modified kernel regression method which exclude depth values having low confidence. Our proposed algorithm guarantees the high quality result in the presence of the camera calibration errors. Experimental comparison with other data fusion techniques shows the superiority of our proposed method.

• Journal of KIISE:Computer Systems and Theory
• /
• v.32 no.11_12
• /
• pp.557-565
• /
• 2005

This paper presents the development and implementation of a zero-copy file transfer mechanism that improves the efficiency of file transfers for PC cluster systems using hardware-based VIA(Virtual Interface Architecture) network adapters. VIA is one of the representative user-level communication interfaces, but because there is no library for file transfer, one copy occurs between kernel buffer and user boilers. Our mechanism presents a file transfer primitive that does not require the file system to be modified and allows the NIC to transfer data from the kernel buffer to the remote node directly without copying. To do this, we have developed a hardware-based VIA network adapter, which supports the PCI 64bit/66MHz bus and Gigabit Ethernet, as a NIC, and implemented a zero-copy file transfer mechanism. The experimental results show that the overhead of data coy and context switching in the sender is greatly reduced and the CPU utilization of the sender is reduced to $30\%\~40\%$ of the VIA send/receive mechanism. We demonstrate the performance of the zero-copy file transfer mechanism experimentally. and compare the results with those from existing file transfer mechanisms.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.31 no.2D
• /
• pp.323-334
• /
• 2011

In this study, we performed the residual geoid modeling using the FFT and LSC methods in context of application of R-R (Remove and Restore) technique as a general technique for gravimetric geoid model in order to propose the effective way of geoid determination in Korea. For this, a number of data compiled for residual geoid modeling by the multi-band spherical FFT method with Stoke's formula and LSC method as known as statistical method. The geometric geoidal heights obtained from 503 GPS/Levelling data were used for inducing the various elements and proper computation process which should be considered for improving the accuracy of residual geoid modeling. Finally, we statistically compared the results of residual geoid heights between FFT and LSC methods and reviewed then the proper way of residual geoid modeling to the region of Korea. As the results of comparison, LSC method is not suitable for residual geoid modeling in Korea due to the noise and lack of gravity observations and the effects of local characteristics, while FFT method by applying Stokes' integral with proper cap size and modified kernel which provides the better accuracy of residual geoid heights up to 10 cm more than those of LSC method.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.15 no.1
• /
• pp.133-138
• /
• 2020

This paper introduces a method for improving accuracy using the first convolution layer, which is not used in most modified CNN(: Convolution Neural Networks). In CNN, such as GoogLeNet and DenseNet, the first convolution layer uses only the traditional methods(3×3 convolutional computation, batch normalization, and activation functions), replacing this with RGB-csb. In addition to the results of preceding studies that can improve accuracy by applying RGB values to feature maps, the accuracy is compared with existing CNN using a limited number of images. The method proposed in this paper shows that the smaller the number of images, the greater the learning accuracy deviation, the more unstable, but the higher the accuracy on average compared to the existing CNN. As the number of images increases, the difference in accuracy between the existing CNN and the proposed method decreases, and the proposed method does not seem to have a significant effect.

• Journal of KIISE:Computer Systems and Theory
• /
• v.34 no.12
• /
• pp.610-617
• /
• 2007

Incremental checkpointing, which is intended to minimize checkpointing overhead, saves only the modified pages of a process. However, the cumulative site of incremental checkpoints increases at a steady rate over time because a number of updated values may be saved for the same page. In this paper, we present a comprehensive overview of Pickpt, a page-level incremental checkpointing facility. Pickpt provides space-efficient techniques aiming to minimizing the use of disk space. For our experiments, the results showed that the use of disk space using Pickpt was significantly reduced, compared with existing incremental checkpointing.

• Journal of IKEEE
• /
• v.20 no.4
• /
• pp.398-411
• /
• 2016

SVM, a machine learning method, is widely used in image processing for it's excellent generalization performance. However, to add other data to the pre-trained data of the system, we need to train the entire system again. This procedure takes a lot of time, especially in embedded environment, and results in low performance of SVM. In this paper, we implemented an SVM trainer and classifier in an FPGA to solve this problem. We parlallelized the repeated operations inside SVM and modified the exponential operations of the kernel function to perform fixed point modelling. We implemented the proposed hardware on Xilinx ZC 706 evaluation board and used TSR algorithm to verify the FPGA result. It takes about 5 seconds for the proposed hardware to train 2,000 data samples and 16.54ms for classification for $1360{\times}800$ resolution in 100MHz frequency, respectively.