Nihon Unisys Ltd.

Tokyo, Japan

Nihon Unisys Ltd.

Tokyo, Japan

Time filter

Source Type

Tsuchie S.,Nihon Unisys Ltd. | Okamoto K.,Daihatsu Motor Co.
CAD Computer Aided Design | Year: 2016

We propose a new method for fitting a high-quality planar curve to styling design data by using a curvature continuous (G2) quadratic B-spline curve. In order to attain G2 continuity of the B-spline curve, we use a non-uniform knot vector, which also enables the curve to be composed of fewer segments as compared to a uniform curve. In our method, control points and the knot vector of the B-spline curve are calculated separately; therefore, we can avoid solving a complicated nonlinear optimization problem. By conducting experiments, we demonstrate that high-quality curves can be generated from both artificial noisy data and real-world scanned data. © 2015 Elsevier Ltd.


Kato K.,Nihon Unisys Ltd. | Hosino T.,Nihon Unisys Ltd.
CCGrid 2010 - 10th IEEE/ACM International Conference on Cluster, Cloud, and Grid Computing | Year: 2010

A recommendation system is a software system to predict customers' unknown preferences from known preferences. In a recommendation system, customers' preferences are encoded into vectors, and finding the nearest vectors to each vector is an essential part. This vector-searching part of the problem is called a k-nearest neighbor problem. We give an effective algorithm to solve this problem on multiple graphics processor units (GPUs). Our algorithm consists of two parts: an N-body problem and a partial sort. For a algorithm of the N-body problem, we applied the idea of a known algorithm for the N-body problem in physics, although another trick is need to overcome the problem of small sized shared memory. For the partial sort, we give a novel GPU algorithm which is effective for small k. In our partial sort algorithm, a heap is accessed in parallel by threads with a low cost of synchronization. Both of these two parts of our algorithm utilize maximal power of coalesced memory access, so that a full bandwidth is achieved. By an experiment, we show that when the size of the problem is large, an implementation of the algorithm on two GPUs runs more than 330 times faster than a single core implementation on a latest CPU. We also show that our algorithm scales well with respect to the number of GPUs. © 2010 IEEE.


Tsuchie S.,Nihon Unisys Ltd.
Visual Computer | Year: 2016

This paper presents a new curve fitting framework for styling design data. Given a data set that represents a filleted-like curve, underlying curves (U-curves) and styling radius corners (SR-corners) are generated by fitting to low curvature parts and highly curved ones, respectively. A set of U-curves are firstly reconstructed as a unique (Formula presented.) composite B-spline curve, and then an SR-corner is reconstructed for each (Formula presented.) corner. This approach guarantees that U-curves can be smoothly connected through convex SR-corners while enabling full editing of the smooth corners up to sharp ones. Compared with existing schemes that naively fit a curve to each part, the proposed framework provides a guiding principle for the generation of curves, which is more suitable for styling design. Experimental results demonstrate that high-quality curves can be generated even from real-world scanned data. © 2016 Springer-Verlag Berlin Heidelberg


Io Y.,Nagoya University | Io Y.,Nihon Unisys Ltd. | Suzuki T.K.,Nagoya University
Astrophysical Journal | Year: 2014

We investigate the formation of hot coronae and vertical outflows in accretion disks by magnetorotational turbulence. We perform local three-dimensional magnetohydrodynamical (MHD) simulations with the vertical stratification by explicitly solving an energy equation with various effective ratios of specific heats, γ. Initially imposed weak vertical magnetic fields are effectively amplified by magnetorotational instability and winding caused by the differential rotation. In the isothermal case (γ = 1), the disk winds are driven mainly by the Poynting flux associated with the MHD turbulence and show quasi-periodic intermittency. In contrast, in the non-isothermal cases with γ ≥ 1.1, the regions above 1-2 scale heights from the midplane are effectively heated to form coronae with temperature ∼50 times the initial value, which are connected to the cooler midplane region through the pressure-balanced transition regions. As a result, the disk winds are driven mainly by the gas pressure, exhibiting more time-steady nature, although the nondimensional time-averaged mass loss rates are similar to that of the isothermal case. Sound-like waves are confined in the cool midplane region in these cases, and the amplitude of the density fluctuations is larger than that of the isothermal case. © 2014. The American Astronomical Society. All rights reserved.


Kato K.,Nihon Unisys Ltd. | Hosino T.,Nihon Unisys Ltd.
Concurrency Computation Practice and Experience | Year: 2012

The recommendation system is a mechanism which automatically recommends items that are likely to be of interest to the user. In the recommendation system, customers' preferences are encoded into vectors, and finding the nearest vectors to each vector is an essential part. This vector-searching part of the problem is called a k-nearest neighbor problem. We give an effective algorithm to solve this problem on multiple graphics processor units (GPUs). Our algorithm consists of two parts: the N-body problem and the partial sort. For an algorithm of the N-body problem, we applied the idea of a known algorithm, although another trick is needed to overcome the problem of small-sized shared memory. For the partial sort, we give a novel GPU algorithm which is effective for small k. In our partial sort algorithm, a heap is accessed in parallel by threads with a low cost of synchronization. We show through an experiment that when the size of the problem is large, an implementation of the algorithm on two GPUs runs more than 330 times faster than a single core implementation on a latest CPU. Copyright © 2011 John Wiley & Sons, Ltd.


Suenaga S.,Nihon Unisys Ltd. | Yoshioka N.,Chiyoda Corporation | Honiden S.,Chiyoda Corporation
Computer Journal | Year: 2011

A wireless sensor network (WSN) consists of spatially distributed nodes that monitor physical conditions. Thus far, WSNs have been constructed for specific applications. However, in the future, system operators may want an existing WSN to deploy simultaneous applications. Existing studies have demonstrated the effectiveness of applying mobile agents to enable simultaneous applications in WSNs. However, two problems make the present form of mobile agents impractical for applications involving cooperation and group migration by mobile agents. The first problem is the lack of a suitable architecture enabling cooperation and interaction between agents, and the second is the lack of an efficient way of achieving group migration by agents in WSNs. To address these problems, we propose an architecture organized by mobile agents that has specific functionalities and that enables group migration by multiple agents. © The Author 2009. Published by Oxford University Press on behalf of The British Computer Society. All rights reserved.


Kato K.,Nihon Unisys Ltd | Hosino T.,Nihon Unisys Ltd
IOP Conference Series: Materials Science and Engineering | Year: 2014

A collaborative filtering predicts customers' unknown preferences from known preferences. In a computation of the collaborative filtering, a singular value decomposition (SVD) is needed to reduce the size of a large scale matrix so that the burden for the next phase computation will be decreased. In this application, SVD means a roughly approximated factorization of a given matrix into smaller sized matrices. Webb (a.k.a. Simon Funk) showed an effective algorithm to compute SVD toward a solution of an open competition called "Netflix Prize". The algorithm utilizes an iterative method so that the error of approximation improves in each step of the iteration. We give a GPU version of Webb's algorithm. Our algorithm is implemented in the CUDA and it is shown to be efficient by an experiment. © 2010 IOP Publishing Ltd.


Tsuchie S.,Nihon Unisys Ltd. | Hosino T.,Nihon Unisys Ltd. | Higashi M.,Toyota Technological Institute
CAD Computer Aided Design | Year: 2014

In order to robustly perform segmentation for industrial design objects measured by a 3-D scanning device, we propose a new method for high-quality vertex clustering on a noisy mesh. Using Student-t mixture model with the variational Bayes approximation, we develop a vertex clustering algorithm in the 9-D space composed of three kinds of principal curvature measures along with vertex position and normal component. The normal component is added, because it well describes the surface-features and is less influenced by noise, and the positional component suppresses redundant clusters due to the normal one. Furthermore, in order to enhance the robustness for noisy data, considering mesh topology as a spatial constraint and letting the vertices in its surroundings belong to the same cluster by diffusion process, we protect generating many small fragments due to noise. We demonstrate effectiveness of our method by applying it to the real-world scanned data. © 2013 Elsevier Ltd. All rights reserved.


Hosino T.,Nihon Unisys Ltd.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2010

Statistical clustering is the method for dividing the given samples by assumed distributions. In high dimensional problems, such as document or image clustering, the direct method is suffered from over-fitting and the curse of the dimensionality. In many cases, we firstly reduce the dimensionality, then apply the clustering algorithm. However these methods neglect the interaction among two processes. In this report, we propose the hierarchical joint distribution of Latent Dirichlet Allocation and Polya Mixture and give the parameter estimation algorithm by Gibbs sampling method. Some benchmarks show the effectiveness of the proposed method. © 2010 Springer-Verlag Berlin Heidelberg.


Tsuchie S.,Nihon Unisys Ltd. | Higashi M.,Toyota Technological Institute
Graphical Models | Year: 2012

In this paper, we propose a novel method for feature-preserving mesh denoising based on the normal tensor framework. We utilize the normal tensor voting directly for the mesh denoising whose eigenvalues and eigenvectors are used for detecting saliency, and introduce an algorithm that updates a vertex by the Laplacian of curvature which minimizes a difference of the curvature in one neighborhood. By connecting the feature saliency with a distance metric in the normal tensor space, our algorithm preserves sharp features more robustly and clearly for noisy mesh data. Comparing our method with the existing ones, we demonstrate the effectiveness of our algorithm against some synthetic noisy data and real-world scanned data. © 2012 Elsevier Inc. All rights reserved.

Loading Nihon Unisys Ltd. collaborators
Loading Nihon Unisys Ltd. collaborators