• The KIPS Transactions:PartA
• /
• v.13A no.2 s.99
• /
• pp.171-176
• /
• 2006

In this paper, we develop energy efficient designs for the Fast Fourier Transform (FFT) on FPGAs. Architectures for FFT on FPGAs are designed by investigating and applying techniques for minimizing the energy dissipation. Architectural parmeters such as degrees of vertical and horizontal parallelism are identified and a design choices. We determine design trade-offs using high-level performance estimation to obtain energy-efficient designs. We implemented a set storage types as parameters, on Xilinx Vertex-II FPGA to verify the estimates. Our designs dissipate 57% to 78% less energy than the optimized designs from the Xilinx library. In terms of a comprehensive metric such as EAT (Energy-Area-Time), out designs offer performance improvements of 3-13x over the Xilinx designs.

• Journal of KIISE:Computer Systems and Theory
• /
• v.31 no.5_6
• /
• pp.288-296
• /
• 2004

This paper proposes a new mesh simplification algorithm using differential error metric. Many simplification algorithms make use of a distance error metric, but it is hard to measure an accurate geometric error for the high-curvature region even though it has a small distance error measured in distance error metric. This paper proposes a new differential error metric that results in unifying a distance metric and its first and second order differentials, which become tangent vector and curvature metric. Since discrete surfaces may be considered as piecewise linear approximation of unknown smooth surfaces, theses differentials can be estimated and we can construct new concept of differential error metric for discrete surfaces with them. For our simplification algorithm based on iterative edge collapses, this differential error metric can assign the new vertex position maintaining the geometry of an original appearance. In this paper, we clearly show that our simplified results have better quality and smaller geometry error than others.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.1
• /
• pp.110-114
• /
• 2010

We describe a new method for text categorization using TextRank algorithm. Text categorization is a problem that over one pre-defined categories are assigned to a text document. TextRank algorithm is a graph-based ranking algorithm. If we consider that each word is a vertex, and co-occurrence of two adjacent words is a edge, we can get a graph from a document. After that, we find important words using TextRank algorithm from the graph and make feature which are pairs of words which are each important word and a word adjacent to the important word. We use classifiers: SVM, Na$\ddot{i}$ve Bayesian classifier, Maximum Entropy Model, and k-NN classifier. We use non-cross-posted version of 20 Newsgroups data set. In consequence, we had an improved performance in whole classifiers, and the result tells that is a possibility of TextRank algorithm in text categorization.

• Journal of KIISE:Computer Systems and Theory
• /
• v.32 no.8
• /
• pp.418-425
• /
• 2005

3D surface reconstruction is to make the original geometry of 3D objects from 2D geometric information. Barequet's algorithm is well known and most widely used in surface reconstruction. It tiles matched regions first, then triangulates clefts using dynamic programming. However it takes considerably long processing time while manipulating complex model. Our method tiles a simple region that does not have branches along minimally distant vertex pairs at once. When there are branches, our method divides contour lines into a simple region and clefts. We propose a fast and simple method that calculates medial axes using a minimum distance in cleft region. Experimental results show that our method can produce accurate models than the previous method within short time.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2009.10a
• /
• pp.397-400
• /
• 2009

This paper describes ASIC design of Multimedia application SoC platform based RISC processor with BTB(Branch Target Buffer). For performance enhancement of platform, we use a simple branch prediction scheme, BTB structure, that stores a target address for branch instruction to remove pipeline harzard. Also, the platform includes a number of peripheral such as VGA controller, AC97 controller, UART controller, SRAM interface and Debug interface. The platform is designed and verified on a Xilinx VERTEX-4 FPGA using a number of test programs for functional tests and timing constraints. Finally, the platform is implemented into a single ASIC chip which can be operated at 100MHz clock frequency using the Chartered 0.18um process. As a result of performance estimation, the proposed platform shows about 5~9% performance improvement in comparison with the previous SoC Platform.

• Journal of the Korea Society of Computer and Information
• /
• v.20 no.1
• /
• pp.85-92
• /
• 2015

As computer game maps get more elaborate, path-finding by using $A^*$ algorithm in grid-based game maps becomes bottlenecks of the overall game performance. It is because the search space becomes large as the number of nodes increases with detailed representation in cells. In this paper we propose an efficient pathfinding method in which the computer game maps in a regular grid is converted into the polygon-based representation of the list of vertices and then the visibility information about vertices of polygons can be utilized. The conversion to the polygon-based map does not give any effect to the real-time query process because it is preprocessed offline. The number of visited nodes during search can be reduced dramatically by designing heuristics using visibility information of vertices that make the accuracy of the estimation enhanced. Through simulations, we show that the proposed methods reduce the search space and the search time effectively while maintaining the advantages of the grid-based method.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.23 no.6
• /
• pp.659-665
• /
• 2019

Various placards are attached to the surface of a container depending on the risk of the cargo loaded. Containers with dangerous goods should be managed separately from ordinary containers. Therefore, as part of the port automation system, there is a demand for automatic recognition of placards. In this paper, proposed is a system that automatically extracts the placard area based on the shape features of the placard and recognizes the contents in it. Various distortions can be caused by the surface curvature of the container, therefore, attention should be paid to the area extraction and recognition process. The proposed system can automatically extract the region of interest and recognize the placard using the feature that the placard is diamond shaped and the class number is written just above the lower vertex. When the proposed system is applied to real images, the placard can be recognized without error, and the used techniques can be applied to various image analysis systems.

• The Journal of the Korea Contents Association
• /
• v.19 no.2
• /
• pp.186-193
• /
• 2019

This article compares the weighted geometric median with the centroid, from the question why they use the centroid when they would find the single facility location(the weighted geometric median) which minimize the sum of weighted Euclidean distances in some text books and papers. Firstly, we show that the demand point whose volume of demand exceeds the half of total demand is the weighted geometric median differently from the centroid, and we examine the weighed geometric median when every demand point is located on a line. Meanwhile, we could simply see that the geometric median and the centroid are coincident in the special case when every demand point is located at a vertex of a regular polygon, and every volume of demand is equal. Furthermore, the geometric medians of convex tetragons could be simply attained unlike triangles.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.22 no.11
• /
• pp.1538-1543
• /
• 2018

For n pairs of points (a,b) on a circle, the line segment to connect two points is called a chord. These chords define a new graph G. Each chord corresponds to a vertex of G, and if two chords intersect, the two vertices corresponding to them are connected by an edge. This makes a graph, called by a circle graph. In this paper, we deal with the problem to find the connected components of a circle graph. The connected component of a graph G is a maximal subgraph H such that any two vertices in H can be connected by a path. When the adjacent matrix of G is given, the problem to find them can be solved by either the depth-first search or the breadth-first search. But when only the information for the chords is given as an input, it takes ${\Omega}(n^2)$ time to obtain the adjacent matrix. In this paper, we do not make the adjacent matrix and develop an $O(n{\log}^2n)$ algorithm for the problem.

• Journal of Biomedical Engineering Research
• /
• v.39 no.6
• /
• pp.237-242
• /
• 2018

$^{18}F$-fludeoxyglucose PET (FDG-PET) can help finding an abnormal metabolic activity in brain. In this study, we evaluated an efficiency of volume- and cortical surface-based analysis which were used to determine whether standardized uptake value ratio (SUVR) of FDG-PET was different among Alzheimer's disease (AD), mild cognitive impairment (MCI), and healthy control (HC). Each PET image was rigidly co-registered to the corresponding magnetic resonance imaging (MRI) using mutual information. All voxels of the co-registered PET images were divided by the mean FDG uptake of the cerebellum cortex which was thresholded by partial volume effect (>0.9). Also, the SUVR value of each vertex was linearly interpolated from volumetric SUVR image which was thresholded by gray matter partial volume effect (>0.1). Lobar mean values were calculated from both volume- and cortical surface-based SUVRs. Statistical analysis was conducted to compare two measures for AD, MCI and HC groups. Even though the results of volume (SUVR_vol) and cortical surface-based SUVR (SUVR_surf) analysis were not significantly different from each other, the latter would be better for detecting group differences in SUVR of PET.