• IEIE Transactions on Smart Processing and Computing
• /
• v.4 no.1
• /
• pp.10-15
• /
• 2015

This paper proposes the parallel implementation of a self-similarity based image SR (super-resolution) algorithm using OpenCL. The SR algorithm requires tremendous computations to search for a similar patch. This becomes a bottleneck for the real-time conversion from a FHD image to UHD. Therefore, it is imperative to accelerate the processing speed of SR algorithms. For parallelization, the SR process is divided into several kernels, and memory optimization is performed. In addition, two GPUs are used for further acceleration. The experimental results shows that a GPGPU implementation can speed up over 140 times compared to a single-core CPU. Furthermore, it was confirmed experimentally that utilizing two GPUs can speed up the execution time proportionally, up to 277 times.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.6
• /
• pp.2648-2668
• /
• 2016

Modern mobile devices are equipped with various accelerated processing units to handle computationally intensive applications; therefore, Open Computing Language (OpenCL) has been proposed to fully take advantage of the computational power in heterogeneous systems. This article introduces a parallel software decoder of Low Density Parity Check (LDPC) codes on an embedded heterogeneous platform using an OpenCL framework. The LDPC code is one of the most popular and strongest error correcting codes for mobile communication systems. Each step of LDPC decoding has different parallelization characteristics. In the proposed LDPC decoder, steps suitable for task-level parallelization are executed on the multi-core central processing unit (CPU), and steps suitable for data-level parallelization are processed by the graphics processing unit (GPU). To improve the performance of OpenCL kernels for LDPC decoding operations, explicit thread scheduling, vectorization, and effective data transfer techniques are applied. The proposed LDPC decoder achieves high performance and high power efficiency by using heterogeneous multi-core processors on a unified computing framework.

• Journal of Korea Game Society
• /
• v.18 no.2
• /
• pp.59-68
• /
• 2018

A visual novel, a type of digital content combining novel and game, was designed as hypermedia which includes various media, for instance, image, sound, and video. The narrative of visual novel contains multimedia elements and highlights multi-scenario structures based on hypertext. Prior to creating visual novel scenario, a writer should thoroughly consider two main points. First of all, the scenario should cover the multimedia features of visual novel. Secondly, the scenario should be structured as a multi-scenario providing with performativity to players for interaction with readers. This study examines the influence of changes in platforms on visual novel scenario adjusted to mobile environment. It would provide an opportunity to understand how storytelling changes and adapts to rapid transformation of digital media contents.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.11 no.6
• /
• pp.454-458
• /
• 2001

Based on a fuzzy system representation of gray scale images, we derive and edge detection algorithm whose convolution kernel is different from the known kernels such as those of Robert's Prewitt's or Sobel's gradient. Our fuzzy system representation is an exact representation of the bicubic spline function which represents the gray scale image approximately. Hence the fuzzy system is a continuous function and it provides a natural way to define the gradient and the Laplacian operator. We show that the gradient at grid points can be evaluated by taking the convolution of the image with a 3$\times$3 kernel. We also that our gradient coupled with the approximate value of the continuous function generates an edge detection method which creates edge images clearer than those by other methods. A few examples of applying our methods are included.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.25 no.1
• /
• pp.15-21
• /
• 2015

In support vector machine, the values of parameters included in kernels affect strongly generalization ability. It is often difficult to determine appropriate values of those parameters in advance. It has been observed through our studies that the burden for deciding the values of those parameters in support vector regression can be reduced by utilizing ensemble learning. However, the straightforward application of the method to large scale problems is too time consuming. In this paper, we propose a method in which the original data set is decomposed into a certain number of sub data set in order to reduce the burden for parameter tuning in support vector regression with large scale data sets and imbalanced data set, particularly.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.47 no.5
• /
• pp.384-389
• /
• 2002

No-tillage (NT) practice for corn production has advantages of reduction of soil erosion and energy conservation. Research on effects of NT for sweet corn or super sweet corn is very limited. Hybrids of sugary (su) and shrunken-2 (sh2) were tested under NT and conventional tillage (CT) practices to investigate plant characters, ear characters, fresh yield, and grain yield. Sugary hybrids were Golden Cross Bantam 70 (GCB70), Sprint, Geumdanok, and Danok3. Shrunken-2 hybrids were BSS9472, Cambella90, GSS9299, Jubilee, KS-Y-65, and Chodangok1. Emergence rates under NT were lower than those under CT for su, while there was no difference between tillage systems for sh2. There were no differences between CT and NT for days to tasseling and silking, plant height, and ear height for both su and sh2. Ear characters such as ear length, number of kernel rows, number of kernels per row, and t100-kernel weight under NT were not significantly different from those under CT. There were no differences between two tillage practice for fresh and grain yield, rather they showed trend of increases under NT practices. Results from this trial indicate that NT practice for both su and sh2 cultivation may be possible to recommend to farmers.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.1
• /
• pp.201-220
• /
• 2020

Trackers, especially long-term (LT) trackers, now have a more complex structure and more intensive computation for nowadays' endless pursuit of high accuracy and robustness. However, computing efficiency of LT trackers cannot meet the real-time requirement in various real application scenarios. Considering heterogeneous CPU-GPU platforms have been more popular than ever, it is a challenge to exploit the computing capacity of heterogeneous platform to improve the efficiency of LT trackers for real-time requirement. This paper focuses on TLD, which is the first LT tracking framework, and proposes an efficient parallel implementation based on OpenCL. In this paper, we firstly make an analysis of the TLD tracker and then optimize the computing intensive kernels, including Fern Feature Extraction, Fern Classification, NCC Calculation, Overlaps Calculation, Positive and Negative Samples Extraction. Experimental results demonstrate that our efficient parallel TLD tracker outperforms the original TLD, achieving the 3.92 speedup on CPU and GPU. Moreover, the parallel TLD tracker can run 52.9 frames per second and meet the real-time requirement.

• Toxicological Research
• /
• v.29 no.2
• /
• pp.143-147
• /
• 2013

Cyanogenic glycosides are HCN-producing phytotoxins; HCN is a powerful and a rapidly acting poison. It is not difficult to find plants containing these compounds in the food supply and/or in medicinal herb collections. The objective of this study was to investigate the distribution of total cyanide in nine genera (Dolichos, Ginkgo, Hordeum, Linum, Phaseolus, Prunus, Phyllostachys, Phytolacca, and Portulaca) of edible plants and the effect of the processing on cyanide concentration. Total cyanide content was measured by ion chromatography following acid hydrolysis and distillation. Kernels of Prunus genus are used medicinally, but they possess the highest level of total cyanide of up to 2259.81 $CN^-$/g dry weight. Trace amounts of cyanogenic compounds were detected in foodstuffs such as mungbeans and bamboo shoots. Currently, except for the WHO guideline for cassava, there is no global standard for the allowed amount of cyanogenic compounds in foodstuffs. However, our data emphasize the need for the guidelines if plants containing cyanogenic glycosidesare to be developed as dietary supplements.

• Proceedings of the Korean Society of Near Infrared Spectroscopy Conference
• /
• 2001.06a
• /
• pp.1021-1021
• /
• 2001

Sample presentation, which means how to set samples to an NIR instrument, is very important in Near Infrared (NIR) Spectroscopy. When sample presentation is not suitable for the samples that you use, very good spectra can not be obtained even if you use a sophisticated NIR instrument. In my presentation, various NIR sample presentations for agricultural products such as intact fruits, single grains, vegetable juice and the others will be explained. In case of peaches with thin peel, the fiber optics of Interactance can be used. However, the fiber optics are not suitable for oranges with relatively thick peel. In this case, transmittance method is useful. As for a small sample such as single grains, a specially designed cell is needed. The cell in transmittance mode has been developed and then applied to single kernels of rice and soybean. In this case we also used the fiber optics. As regards liquid type of sample, a cuvette cell made of quartz in transmittance mode is popular. However, it is time-consuming to wash and dry it. In order to compensate this disadvantage the sample presentation using normal test tubes as sample cells have been developed and applied to milk, rumen juice and urine of a milking cow. An individual test tube can be used for each sample if you use the calibration equation with sample cell compensation. The test tube cell has also been applied to spinach juice for determination of undesirable constituents. It is concluded that sample presentation is most important for NIR Spectroscopy.

• Proceedings of the Korean Society of Near Infrared Spectroscopy Conference
• /
• 2001.06a
• /
• pp.1616-1616
• /
• 2001

To select kernels for breeding that have required constituent content from either naturally distributed samples or artificially mutated ones, it is necessary to process batch samples in a short time. The constituent content of single-kernel grains such as wheat and rice has been determined using conventional bench type NIR instruments; however, it takes a lot of time and effort. Shizuoka Seiki (Fukuroi-city, Japan) and NFRI (National Food Research Institute) of MAFF (Ministry of Agriculture, forestry and Fisheries of Japan) have jointly developed a continuous high-speed single-kernel brown rice sorting machine based on rice protein content. It consists of several sections such as a feeding mechanism, measuring unit, sorting mechanism and controlling PC. The feeding mechanism picks up single-kernel brown rice from the hopper (maximum of 5kg storage capacity) and sends it to the measuring unit. A spectrum of the brown rice is obtained in the measuring unit, which consists of a near-infrared array sensor. The brown rice is then sorted in the sorting mechanism based on its protein content estimated by the controlling PC. In the present study, measuring speed was approximately 500ms for the full spectrum range and overall sorting speed was approximately 2.8s for one kernel. Accuracy of estimation was approximately SEP=0.5% of dry matter protein content for nonglutinous rice.