vol.14 (2001)

Bulletin of Computational Science Research Center
Vol.14, 2001 ISSN 0913-8420 Published March 31, 2001

Wearable Haptic Display for Immersive Virtual Environment
Hisayuki Yamauchi, Kenichi Amemiya, Yutaka Tanaka
PDF   Abstract
Recentry, a number of immersive displays that project virtual environment on large-sized screens have been developed. In applications of virtual reality technology for such large immersive environments, it is important to develop some haptic display that gives a real feedback sensation such as force or tactile to users from the virtual environment. Especially, it is necessary to develop small, light and safety actuators for the wearable haptic display. In this paper, the haptic display system for the immersive virtual environment is proposed and developed. The system consists of a tactile and a force reflecting type wearable haptic displays using pneumatic pressure control. Experiments for grasping virtual objects and recognizing positions and figures of virtual objects in the virtual environment are performed by the wearable haptic displays using pneumatic actuators. It is experimentally verified that the performance of the haptic display sysytem is effective to touch and grasp the virtual objects.
Simulation of Flow in Turbomachinery By Using 2-D Curved Duct (Influence of Aspect Ratio)
Ryuji CHIDA, Takuya KOMATSU, Masataka NAITO, Hoshio TSUJITA, Shimpei MIZUKI
PDF   Abstract
In the passage of a turbomachinery, the secondary flow under the effect of the centrifugal and the Coriolis forces exhibits complex flow pattern, and strongly affect the generation of loss. It is difficult to reveal the relationship between the secondary flow and the loss generation mechanism by using an actual turbomachinery. The most dominant secondary flow is the passage vortex. In the present study, the flow within the two-dimensional curved duct is selected as a fundamental model for the appearance of the passage vortex. The numerical computations are performed for the various conditions, in which the aspect ratio of curved duct are changed.
Analysis of Ester Interchange by Semi-empirical Molecular Orbital Method
Atsushi Kondo, Yosuke Kataoka
PDF   Abstract
We simulated the reaction of ethyl acetate with methanol and searched for a process by program package of semi-empirical molecular orbital calculation, MOPAC. As a result, we confirmed that ethyl acetate reacts with methanol, and then they formmethyl acetate and ethanol. And we found the ratio of eater to alcohol is 1 to 2 in the case of using acid catalyst.
Correlation of Liquid-liquid Equilibria for Binary and Ternary Components with UNIQUAC equation using Marquardt Method Proposed
Hirotake Katayama
PDF   Abstract
UNIQUAC or NRTL equation has been usually used to correlate experimental results of liquid-liquid equilibria for binary and ternary components, however, in deciding the parameters of the equation some particular techniques are needed. This report explained the techniques using Marquardt method, which was modified by the author.
Improvement in Accuracy of reconstruction on Stereo Vision based on Color Segmentation
Takashi Akimoto, Hironao Saito, Yoji Hamano, Masami Iwatsuki
PDF   Abstract
We proposed a new stereo vision method that can reconstruct 3D structures robustly by matching left and right images based on regions given by color segmentation. However, the accuracy of reconstruction is limited to the pixel resolution on this method because the corner detection is based on a line segmentation for region contours. Therefore, this paper proposes an approach to improve the accuracy of corner positions on contours to the subpixel resolution by using a maximum likelihood method.
High Accurate and Complete 3D Reconstruction based on Active Vision
Tomoaki Ohishi, Miyuki Katsumata, Masami Iwatsuki
PDF   Abstract
We proposed a new stereo vision method that can reconstruct 3D structures robustly by matching left and right images based on regions given by color segmentation. However, the acquired 3D map inevitably contains lacking or error parts caused by occlusion and noise. Therefore, this paper proposes an approach to detect and modify the error and lacking parts by fusing multiple 3D maps acquired by the active vision.
Fault Correspondence Detection by Consistency of Disparity of Region Contours on Scanline Stereo Matching
Toyokazu Ogasawara, Yuichi Katase, Masami Iwatsuki
PDF   Abstract
We proposed a new robust method of stereo matching using coarse to fine hierarchical color segmentation of each scanline in stereo images. However, it is difficult to remove fault correspondences in the case of that similar sequences of color segments exist in multiple in the proposed method. Therefore, this paper proposes an approach to detect and modify fault correspondences by taking into account of consistency of disparity of region contours.
3D Reconstruction Based on Separation of Texture and Uniform Color Regions
Shouta Sato, Nobuo Abe, Masami Iwatsuki
PDF   Abstract
This paper proposes 3D reconstruction method using separation of texture and uniform color regions. The proposed method can reconstruct 3D shapes for complicated scenes mixed with texture and uniform color regions by taking account of the correlation, validity of texture and resemblance obtained from GLDM.
Limitation of Search Range and Detection of Occlusion for Correlation Stereo Method
Toshi o Suzuki, Yoshihito Seto, Masami Iwatsuki
PDF   Abstract
In conventional correlation stereo methods, it is difficult that the disparity is obtained with accuracy in occluded parts, since correct corresponding points cannot be found. Therefore, this paper proposes a new stereo method that can limit the range for the matching by segmenting each scanning line in right and left images and detecting same color series pattern of the right and left regions before it is adapted to a correlation method. The fault correspondences can be greatly reduced even in the discontinuous boundary of the depth by this preprocess, since occluded regions are removed.
An efficient learning algorithm for binary neural networks
Masanori Shimada, Toshimichi Saito
PDF   Abstract
This paper proposes a novel efficient geometrical learning algorithm for binary neural networks: a feed-forward neural network having binary inputs and outputs. Our supervised learning algorithm uses flexible and speedy linear separation method using virtual inputs. As compared with conventional algorithm, our algorithm can reduce the number of hidden layers and improve the variation of the connection parameters for complex teacher signal. The learned networks are suited for hardware implementation.
Quasi-analytical Electromagnetic Field Analysis
Yasuyuki Watazawa, Seiji Hayano, Yoshifuru Saito
PDF   Abstract
Any of the numerical methods for electromagnetic field analysis essentially require a subdivision of a problem region. By notifying this subdivision, we have previously proposed a new methodology by applying the analytical solution to each of the discretazed parts. The first trial of this quasi-analytical electromagnetic field computational methodology has made it possible to simulate the complex electromagnetic field distributions not obtainable by the conventional numerical schemes, such as finite elements and boundary elements means. However, our quasi-analytical method encounters some difficulty when analyzing the high frequency electromagnetic field distribution because of the displacement currents. In the present paper, we propose a new quasi-analytical approach taking the displacement currents into account to overcome this difficulty.
Evaluation of Multi-layered Currents by the Least Squares Method
Hiroyuki Takahashi Seiji Hayano Yoshifuru Saito
PDF   Abstract
Modern electronic devices always contain printed circuit boards (PCBs) with multiple layers. If the currents in the PCBs can be visualized without disassembling the electronic devices, the testing and inspection of the devices can be carried out extremely efficiently. This paper proposes a methodology for estimating the current distribution in multi-layered PCBs. The estimation of the current distributions in a PCB from locally measured magnetic fields can always be reduced to solution of an ill-posed inverse problem. This paper reveals that the conventional least squares method give a reasonable solution of our inverse problem. Thus, we succeeded in realizing a highly reliable non-destructive testing methodology for electronic devices.
A study of magnetic sensor signal analysis
Yukiyasu Shigeta, Seiji Hayano, Yoshifuru Saito
PDF   Abstract
We are now exploiting a magnetic sensor, which detects metallic materials, embedded into the ground and concrete walls. Our magnetic sensor is composed of the mechanically vibrating DC magnet having the sensing coils. When there are no metallic materials in the DC vibrating magnetic field, no magnetic field distortion senses no magnetic materials. However, if there is a kind of metallic materials in the DC vibrating magnetic field, then the magnetic field distortion induces a signal at the sensing coils. Analysis of this sensor signal leads to the physical properties of the target metallic material. In the present paper, we propose a method of analysis for the magnetic sensor signals. Based on the physical characteristic value such as a time constant of the electric circuit, our methodology tries to work out an equivalent characteristic value reflecting the physical property of the target metallic material. As an initial experiment, we have carried out an evaluation whether the target metallic material is a magnetic or non-magnetic material. As a result, we have succeeded in cognition of the magnetic and non-magnetic materials. Thus, second stage of our smart magnetic sensor developing is to recognize a datum from the signal database. Furthermore, this method is applied to the space domain signal recognition. As a result, this paper reports that the equivalent characteristic value extracted from our sensor signal is a useful quantity in order to examine the physical nature of the target.
Magnetic Sensor Analysis by Finite Elements
Hironori Kouch, Hisashi Endo, Seiji Hayano, Yoshifuru Saito
PDF   Abstract
Time domain electrical signal represents various physical signals, for example, sound, pressure and so on. In any signals, it is classified into two major signal components. One is a meaningful signal reflecting on a physical cause. The other is the noise caused by various physical and artificial conditions, e.g., electromagnetic inductions when converting the physical signals into electrical quantities. Thereby, extraction of a meaningful time domain signal is one of the most important processes in the sensor signal processing. In the present paper, at first, we evaluate the finite elements solution of a magnetic sensor in order to obtain the noise free sensor signal components. Second, we carry out the practical sensor signal measurements. Third, we compare them to estimate the noise signal components. Thus, it is possible to exact only the meaningful sensor signals theoretically. To search for the noise cause, we visualize the magnetic fields of the sensor and check up the electromagnetic disturbances caused by the sensor target. As a result, it is found that the major electromagnetic noise components are caused by the mechanical accuracy of the sensor structure not by the environmental electromagnetic inductions.
Estimation of Quasi-3D Current Distributions in a Square Cubic Box
Daishiro Sekijima, Seiji Hayano, Yoshifuru Saito
PDF   Abstract
We have studied the visualization of the current distributions from the locally measured magnetic fields. In order to visualize the current distributions, it is intrinsically reduced into solving an ill-posed inverse problem. In most cases, the solutions of the ill-posed inverse problems tend to increase the noise. In this paper, we propose one of the methodologies of the noise processing. At first, we apply the generalized vector sampled pattern matching (GVSPM in short) method to visualizing the quasi-three dimensional current vector distributions. Second, in order to extract the major current vector distributions, we apply the 3D wavelets multi-resolution analysis to the noisy current vectors. Consequently, we have succeeded in visualizing the quasi-3D current vector distributions in a square cubic box.
Identifying the Coil Elements Along With the Eigen Pattern of Magnetic fields
Kenichi Wakabayashi, Seiji Hayano, Yoshifuru Saito
PDF   Abstract
A new concept is introduced to extract the essential and distinct characteristics of the magnetic field distributions. A key idea is to calculate an eigen pattern, which can represent the characteristics of magnetic fields independent of their resolutions and space positions. The eigen pattern of a magnetic field distribution is obtained by projecting the three orthogonal magnetic field vector components at each of the space coordinates onto the x, y and z magnetic field components coordinate system. Thereby, this projection is possible to remove the space coordinate information, and extract the essential vector characteristics of the target magnetic field distribution. In this paper, we have succeeded in estimating the positions of each coil element from compound model along with eigen pattern.
Visualization of the Electromagnetic Field Vectors
Akira Ema, Seiji Hayano, Yoshifuru Saito
PDF   Abstract
We propose the new magnetic field control methodology based on the inverse approach. As an initial test trial, we try to realize the desired magnetic field distribution by the single side planner current carrying coils. Then, we employ the least squares and generalized vector sampled pattern matching (GVSPM, in short) methods as the inverse approach. As a result, it has been confirmed that our approach gives the physically realizable results. By means of the simulation as well as experiment, the validity of our method has been verified.
Magnetic sensor signal processing and cognition
Isamu Senoo, Seiji Hayano, Yoshifuru Saito
PDF   Abstract
Cognition of the time domain signals has been mostly depended on the experiences of the skillful observers. We have proposed a new methodology in order to carry out for human voice cognition [1]. Since the human voice depends on each of the persons even if the same word pronouncing, then it is essentially accompanying the data based learning process. To overcome this difficulty, we have proposed the image processing methodology for which the human voice signals are converted into the three-dimensional monochrome images. The human voice time domain signals are converted into the three-dimensional image data by means of the modified Lissajous diagram. In the present paper, we apply the image cognition technology of the human voice to the cognition of magnetic sensor signals. At first, the time domain signals are converted into the three-dimensional monochrome images, which construct the signal database system. Secondly, when we measure a time domain signal, this signal is also converted into the three-dimensional image. This three-dimensional image becomes an input vector of a least square system. Least squares solution gives a composite signal as a linearly combined database signals. Extracting the most dominant term from the least squares solution reveals the cognized signal. Thus, we have succeeded in the time domain magnetic sensor signal cognition by means of the image cognitive technology.
A study of Planar Inductor Design
Tomokazu Naruta, Hisashi Endo, Seiji Hayano, Yoshifuru Sai
PDF   Abstract
In order to design a planar inductor, this paper carries out several axisymmetrical magnetic field simulations fully taking the magnetic hysteretic properties of ferromagnetic core materials as well as open boundary condition into account. Finite element method is applied to evaluating the magnetic fields, reactive and dissipative energies in a planar inductor. Because of the nature of magnetic fields, it is essentially taken into account the open boundary effects to evaluate the exact characteristics of planar inductor. Since the finite elements is one of the methodologies to solving for the partial differential equations, and then it has been difficult to take the open boundary effects into account. However, this paper removes this difficulty by employing the strategic dual image (SDI, in short) method. Another difficulty is caused by the hysteretic magnetization characteristics of magnetic core materials. Applying the Chua type magnetization model to represent the hysteretic property also removes this difficulty. As a result, it is revealed that the frequency characteristic of core magnetic materials as well as shape dominate major characteristics of the magnetic fields, reactive and dissipative energies in the inductors.
Evaluation of Magnetic Materials by Means of Image Differential Equations
Hisashi Endo, Seiji Hayano, Yoshifuru Saito, Masahiro Fujikura, Chikara Kaido
PDF   Abstract
This paper proposes an image processing approach for analyzing the magnetic domain behavior by means of image Helmholtz equation. According to our methodology, any magnetized domain states can be visualized as one of the animation frames from the finite number of scanning electron microscope (SEM) images. The microscopic magnetic domain behavior as well as macroscopic magnetization process are analyzed based on our image Helmholtz equation described in this paper. The state transition matrices derived from a series of distinct SEM images correspond to the Preisach density functions. Thereby, the magnetic domain dynamics can be visualized as one of the animations. In this paper, we apply our methodology to magnetic domain SEM images of a grain-oriented electrical steel. As a result, we have succeeded in clarifying and visualizing the iron loss generation mechanism.
Magnetic Circuit Analysis based on the Finite Elements -Generalization to the Quasi-Three-Dimensional Problem-
Hayato Yabunami, Seiji Hayano, Yoshifuru Saito
PDF   Abstract
As is well known, conventional magnetic circuit theory makes it possible to analyze the three-dimensional magnetic fields, even though its theoretical background is based on the engineering experiences and its accuracy essentially depends on the assumed magnetic circuits. On the other side, because of the widely spreading use of the digital computers, most of the magnetic field analysis is being carried out by finite elements. The finite elements have a firm theoretical background but it is too expensive to implement the three-dimensional analysis. In order to carry out the three-dimensional magnetic field analysis in a most efficient manner, we propose the theory of modern magnetic circuits(TMMC, in short) based on finite elements discretization. This makes it possible to implement the three-dimensional magnetic field computation in a quite efficient manner similar to those of the conventional magnetic circuit theory. In the present paper, we generalize the two-dimensional theory of modern magnetic circuits to the quasi-three-dimensional magnetic circuits. A simple example demonstrates not only the usefulness but also validity of our new methodology.
Study on analitical image correction of attenuation and scattering effect in SPECT
Norihiro Watanabe, Koichi Ogawa
PDF   Abstract
Attenuation and scattering of gamma rays and the collimator aperture cause degradation of the image quality in single photon emission computed tomography(SPECT). Several methods were proposed for compensating the attenuation of gamma rays and the aperture effect of the collimator. This paper focuses on the analitical method for correcting the attenuation effect and aperture effect. The methods used here are Bellini’s method and Xia’s method. The performance of these correction methods to the noisy projection data were discussed by using the results of computer simulation.
Detection and Classification of Nonstationarity in Time Series
Masaaki Endo, Hirochisa Mizuta, Kazuio Yana
PDF   Abstract
A method of deteating and classifying nonstationarity in observed time series is proposed. The method enables dynamia unsupervised classification by the combination of self-organizing maps (SOM) and the lattice filter. The method has been validated by a speech signal analysis. Japanese vowels hiave been successfully classified by thie proposed method. The method may be useful for the analysis of various nonstationary signals suoh as speech signals, local area network traffic changes or biosignals.
On Selfl-similarity ofl the SNMF Time Series of an Information Network
Hideaki Shino, Tomofumi Ohkuabo, Hitoshi Akimoto, Kazuo Yana
PDF   Abstract
Network traffic is currently exploding worldwide due to the rapid development of broadband network technolegy and communication demand such as WWW browsing. To clarify thle statistical properties of thle network traffic is important for designing an efficient network system. This paper confirmed thle presence of self-similarity in a university LAN traffic i.e. SNMP time series by thle R/S analysis. Thle evidence will be useful in thle prediction of network traffic and evaluation of network devices.
Application of Natural Element Method to the Solid Mechanics
Satoshi MATSUO, Hiroshi TAKEDA
PDF   Abstract
In this paper, we assess the potential and capabilities of a recently developed numerical method-coined as Natural Element Method. The interpolation scheme used in NEM is known as natural neighber interpolation. The application of NEM to the solution of elliptic boundary value problems in solid mechanics, governed by the equation of equilibrium in elastostatics, is explored. In the application of NEM to the solution of the equilibrium equation of elastostatics, a displacement-based Galerkin implementation is used, and the trial and test function are constructed using natural neighber interpolation.
A Feasibility Study on Application to Detailed Digital Information for Runoff Analysis
Yuki MATSUURA, Koji KODERA, Yasumiti OKA
PDF   Abstract
In analyzing stormwater runoff by distributed runoff model, representation of spatial distribution of geographical information is required. The purpose of this study is to estimate the applicability of Detailed Digital Information to runoff model. There remained some problems in the evaluation method for runoff coefficient in each land use and modeling of runoff control facilities in urbanized basin. However, the obtained result shows that Detailed Digital Information can be effectively applicable to distributed runoff model.
An analysis of progressing crack by using hybrid-type penalty method
Hirohisa Ohki, Norio Takeuchi, Morito Kusabuka, Hiroshi Takeda
PDF   Abstract
In this paper, material non-linear analysis considering tensile fracture was developed from the viewpoint of hybrid displacement model. In this model, the concept of the spring of RBSM was introduced in Lagragian multiplier. In addition, compatibility of the displacement on the intersection boundary was approximately introduced using the penalty as a spring constant. As a result of the slip analysis, the slip lines and the collapse load were equal to RBSM. The equivalent tensile collapsing load with FEM was obtained when it carried out analysis of the bending problem for the plain concrete beam.
An analysis of seepage flow problem by using hybrid-type penalty method
Chiaki Ishigaki, Norio Takeuchi, Morito Kusabuka, Hiroshi Takeda
PDF   Abstract
In this paper, new discrete method which applied the approach of the hybrid model to the seepage flow problems are developed. First, the formulation which introduced the potential continuity by penalty in general weighting residual procedure is shown. Compatibility of the potential on the sub-domain boundary edge is approximately introduced using the penalty. Next, the Taylor’s series expansion is used for potential function. It is possible that this model uses the subdomain of the arbitrary shape, because independent potential field is assumed for each sub-domain. The equivalent solution with the analytical solution when it examined the accuracy of numerical solution using this model is obtained.
Simulation for traveling path of snow avalanche
Norio Takeuchi
PDF   Abstract
In this paper, in order to estimate the run-out distance of snow avalanche, its traveling path and spreading width are studied from view point of kinematics as well as dynamics. The influence range of the snow avalanche is discussed with a dynamic analysis of the snow avalanche on three-dimensional topography based on the model of Perla and others. A simplified simulation technique for the spread of the snow avalanche with an application of the analysis is proposed. Some numerical example are shown in the ski resort with a model topography.
Consolidation Yield Function of Saturated Clay and 3-D Consolidation Analysis
Masamichi Ohashi, Morito Kusabuka, Norio Takeuchi
PDF   Abstract
An equation of the linear consolidation theory for saturated and porous elastic body was derived by Biot. However, the equation includes assumptions that are different from real consolidation phenomena The purpose of this study is to develop a 3-D elasto-plastic FEM program for consolidation problem, and evaluate and examine a practicability. The results show a difference between linear elastic analysis and elasto-plastic analysis to disappearance process of excess porewater pressure and settlement of consolidation, etc.
Effectiveness of Object-Oriented Design on Finite Element Method Programming
PDF   Abstract
A number of objects in civil engineering are affected by diverse conditions therefore those phenomena are really complicated in structure. That has brought with it following increase necessity of programming alteration and degree of difficulty modification. In this study, Finite Element Analysis was modeled by Uniformed Modeling Language on Object-Oriented Design and programmed by Java. Model the target by all-purpose language, then end user and programmer are able to catch the system objectively and that is easy to understand rightly, smoothly each other. On programming phase, it has discovered that following faculties, Matrix operation class library for FEM, useful of polymorphism on inheritance as a code recycle, remote development and handling on networking, and so on those couldn’t achieve until now, help the way brake through. The result from comparison of Procedure-Oriented and Object-Oriented is that there were signs of effectively to use Object-Oriented and enough motive to adopt it.
Introducing Object Oriented Design to Finite Element Method Program and implementing in Java
Msashi SATO, Morito KUSABUKA, Norio TAKEUCHI, Hiroshi TAKEDA
PDF   Abstract
This study aims at discussing design, implementaion, and usefulness in the Object Oriented Program. The Finite Element Method Program was designed by The Unified Modeling Language and was implemented in Java. The program was compared with another program inplemented inFORTRAN. As a result, The program had nearly equality performance. The Finite Element Method Program by Object Oriented has enough usefulness in technological computation.
Nonlinear Elasticity And Permeability Constitutive Equation of Saturated Clay And The Three-Dimensional Consolidation Analysis
Daisuke Itakura, Morito Kusabuka, Norio Takeuchi
PDF   Abstract
In the numerical analysis of the consolidation governing equation , a material parameter treated as constant in process of consolidation progress. However, it is considered that the governing equation should consider non-linear of material. For the purpose of relating the change of the permeability in the progression of the consolidation to the void ratio , the consolidation and permeability tests is carried out to normal consolidated clay. The non-linear constitutive relation between permeability and void raitio and Young’s modulus defined a function of effective hydrostatic stress was applied to three-dimensional consolidation program. Compare the non-linear analysis results with the linear was examined.
Numerical Integrations and Differentiations of Seismic Waves by means of Continuous Wavelet Transform
Yuko Oguchi
PDF   Abstract
Numerical integrations or differentiations are not so difficult. In seismology, the neccessity of calculations of displacements and/or velocities from acceleration records of ground motions has been occured frequently. These calculations are usually done in the time domain to have not been derived to accurate results due to the high and lo frequency noises contained in the original data. The autor has noticed that the Foureir transform of Mexican Hat wavelet has ω2 factor and this term will eliminate the low frequency part of data. The high frequency noise are easily omitted from wavelet transform. So, he applied the continuous Mexican Hat wavelet transform to calculate integrations and differentiations of them and have show this method will be quite effective to some groud vibration data.
A Study of Digital Watermark
Norimichi Ishida
PDF   Abstract
Recently, The internet is exponentially developed on digital society. As result, We can get easily literacy work without permission from author. The infringement of copyright is carry weight worthy of remarkable problem in net society. watermark is one of the recently remarkable technology for preventing form infringement of the copyrights for digital data of the original content. We have presented a one of method of digital watermark for the purpose of copyrights protection by frequency domain through wavelet transform.
Vector Fields Animation by Wavelets Transform
Sawa Matsuyama, Yuko Oguchi, Shiho Matsuyama, Yoshifuru Saito
PDF   Abstract
Major applications of the wavelet transform have been focused on the waveform analysis and image data compression. One of the distinguished properties of the wavelet transform is that the major dominant factors can be extracted from the data. We have applied this property to the data compression and reducing the noise data. In the present paper, we have tried to shrink and enlarge the wind vector image data by the three-dimensional discrete wavelets transform. Several examples (vector fields animation) demonstrate the usefulness of our new method to work out the graphical communication tools.