Showing papers presented at "Virtual Environments, Human-Computer Interfaces and Measurement Systems in 2004"

Segmentation and quantification of brain tumor

Abstract: Nowadays, the inside situation could be obtained by medical machines, such as computed tomography (CT), magnetic resonance imaging (MRI) scanner, etc. These noninvasive diagnosis means increase the precision of the diagnoses, at the same time decrease the pain of the patients. Brain tumor diagnoses benefit from these devises very much. In the brain MR image, the tumor is shown clearly. For the treatment, the physician also needs the quantification of the tumor area. This requires the abnormal part in the image to be segmented accurately; afterward the segmented area can be measured. This task could not be handled by hands totally. Vie computer can give great help during this procedure. In this paper we introduce one system to perform the task mentioned above. In this system, the brain tumor MR image is segmented semi-automatically. One deformable model-based method is adapted in the system. And by the graphic user interface, the segmentation can be intervened by user interactively at real time.

On the effectiveness of MFCCs and their statistical distribution properties in speaker identification

Abstract: This paper presents a study on the effectiveness of mel-frequency cepstrum coefficients (MFCCs) and some of their statistical distribution properties (skewness, kurtosis, standard deviation) as the features for text-dependent speaker identification. Multi-layer neural network with backpropagation learning algorithm is used here as the classification tool. The MFCCs representing the speaker characteristics of a speech segment are computed by nonlinear filterbank analysis and discrete cosine transform. The speaker identification efficiency and the convergence speed of the neural network are investigated for different combinations of the proposed features. The result shows that the first MFCC degrades the identification competence and statistical distribution parameters enhance the training speed of the neural network.

Realtime 3D position detection of human pupil

Abstract: A real time 3D pupil position detection system necessary for non-contact and remote eye gaze detection allowing head displacement was developed. So far, methodology had been proposed that estimates 3D positions of eyes from their 2D positions detected from the face images captured by two stereo cameras. However, using eye position detection is not prospective for precise and stable eye gaze detection. In the method proposed in the present study, several near-infrared LEDs were arranged around the apertures of two cameras. These two sets of light sources were alternately switched on and off synchronously with the even/odd signal. After differentiating the bright and dark pupil images obtained from each of the two cameras, the center coordinates of the two pupils were detected. An image processing algorithm for precise and stable pupil detection, including the pupil detection method utilizing a blink, is described. In one of the experiments, the 3D position of the pupil of one eye was measured by the two stereo cameras. Another narrow camera, which detects the centers of both pupil and corneal reflection images necessary for eye gaze direction determination, placed on a pan-tile unit could be automatically directed toward the 3D pupil position.

DPSD algorithm for AC magnetic tracking system

Abstract: 6 DOF (degree-of-freedom) tracking system is one of the key device to realize the sense of immersion and human computer interaction in a VR or AR (Virtual or Augmented Reality) system. This paper presents the design of an AC magnetic tracking system and a DPSD (digital phase sensitive detector) algorithm to extract the received signals. The proposed magnetic tracking system is composed of 3-axis orthogonal magnetic sensor, 3-axis orthogonal magnetic transmitter, 2-axis accelerometers, data acquisition and processing system etc. After obtaining the orientation of the receiver by measuring the Earth magnetic and gravity field with the DC output of magnetic sensors and accelerometers, the position of the receiver can be calculated from the received AC magnetic field generated by the magnetic transmitter. The performance of the existing DPSD algorithm when being used to extract the received AM (amplitude modulation) signals is studied and a new DPSD algorithm as well as its performance are presented. Experimental results show that the proposed algorithm can be used to measure the AM signal embedded in background noise and the feasibility of its application in the AC magnetic tracking system.

A pocket-PC based navigational aid for blind individuals

Abstract: This paper describes a pocket-PC based electronic travel aid (ETA) that helps a blind individual navigate through indoor environments. The system detects surrounding obstacles using ultrasonic range sensors and the travel direction using an electronic compass. The acquired information is processed by a pocket-PC to generate a virtual acoustic environment where nearby obstacles are recognizable to the user. This virtual environment is played back through stereo headphones, so that the user can perceive surrounding obstacles and the direction of the Earth's magnetic North, using spatialized 3D sounds. The paper describes the instrumental and computational aspects of the design and presents the results, demonstrating the improvement in blind travel achieved with the system.

Hand posture classification using wavelet moment invariant

Abstract: In this paper, we present a wavelet moment invariants for classification a small change in rotation and subtle difference of hand posture causes misclassifying to other postures. The method combined zernike moment to capture global features and wavelet moment to differentiate between subtle variations in description can be utilized at the same time. Then, a fuzzy classification algorithm is used to classify hand posture. The classification rate obtained is 72% with of Thai sign language.

A 3D-anthropometric-muscle-based active appearance model

Abstract: This paper describes a novel method for modeling the shape and appearance of human faces in 3D using a constrained 3D active appearance model (AAM). The method uses a generic 3D wireframe model of the face, based on two sets of controls: the anatomically motivated muscle actuators to model facial expressions and statistically based anthropometrical controls to model different facial types (3D-anthropometric-muscle-based-model, 3D-AMBM). This allows explaining a facial image in terms of a controlled model parameter set, hence providing a natural and constrained basis for face segmentation and analysis. The generated face models are consequently simpler and less memory intensive compared to the classical appearance based models. Additionally, our method achieves accurate fitting results by constraining solutions to be valid instances of a face model.

An interface for semantic browsing of an images database

Abstract: In this paper, the authors propose an interactive graphical interface suitable for semantic browsing of an image database. This interface allows user to explore the effects of features integration, multi-image queries and relevance feedback in enhancing the performance of an image retrieval system. Single image query is the retrieval model that is traditionally used, while the authors propose the implementation of a new methodology consisting of multiple query images. Features integration and feedback improve the performance of the proposed system significantly. These techniques require an interactive interface which is able to gather information from user and to show him the view of the image database closest to his requirements.

Biometric authentication using finger and palmar creases

Abstract: A practical method for a non-contacting and real-time authentication is proposed. The finger geometry and feature extraction of the palmar flexion creases are integrated into a small number of discrete points. For a video image of either palm, a palm placed freely facing toward a video camera is acquired. The fingers are brought together and the palm is straightened out without any constraints. The discrete feature points involve intersection points of the three finger (digital) flexion creases on the four finger skeletal lines and the intersection points of the major palmar flexion creases or prominent creases on the extended finger skeletal lines and the orientations of the creases at the points. The matching so far results are perfect for about 500 palm samples from 50 subjects. This discrete point processing, requiring no time-consumptive palm print image analysis and requiring less than one second processing time, will contribute to a non-contacting, real-time and reliable authentication.

A study on an efficient pre-resolution method for embedded Java system

Abstract: Java has been recognized as an attractive language and platform to program embedded systems. It's not appropriate to apply class files in standard Java to embedded Java with the memory and processor limitation of embedded Java. In this paper, we propose an efficient method that can reduce execution time by using pre-resolution to convert symbolic references information to real addresses before executing of bytecode in target device. We examined 13 class files to know the sizes of the constant pool elements in the class file. Our pre-resolution, it reduces the overall memory footprint to about 87% of their original size. Additionally, it reduces a reference count of memory.

A perceptive multirobot system for monitoring electro-magnetic fields

Abstract: Sensor networks are distributed systems that detect phenomena to produce detailed environmental assessments. The use of multirobot systems for carrying sensors around the environment represents a solution that has recently received considerable attention. In this paper we present an architecture for a multirobot system that performs tasks related to environmental perception; in particular, it is devoted to monitoring electro-magnetic fields.

Improving robot path planning efficiency with probabilistic virtual environment models

Abstract: Probabilistic multiresolution occupancy grid modeling has recently been developed to map both 2D and 3D cluttered spaces. These models can be used to provide an enhanced representation of the cluttering state of space in a robot workspace. As a result, they reveal to be promising tools to improve classical potential field based robot path planning strategies. These approaches rely on a combination of repulsive and attractive potential fields to attract the robot toward a given goal while ensuring safe distance from the obstacles. This paper proposes an new approach to directly compute repulsive and attractive potential fields from the probabilistic occupancy models without the need for distance tables or wave propagation. Experimentation revealed that multiresolution probabilistic models encoded as quadtrees or octrees significantly reduce processing time and speed up robot operation.

Development of man machine interface software for an industrial robot

Abstract: This paper describes the design and implementation of a man machine interface software for the ABB IRB-2000 industrial robot. The main aim of this software is to provide features which would facilitate on-line/off-line programming of the robot. The software would provide a three dimensional graphical simulation of the robot and its environment, which also allows loading custom designed objects into the simulated environment. The software also offers collision detection capability between graphically simulated objects, an interpreter/editor to interpret/write robot programs in BASIC like language enhanced with robot commands. The facility to control the actual robot through the computer's serial communication port is also provided.

Implementation of a virtual environment for interacting with a coordinate measuring machine

Abstract: This paper focuses on implementation issues of a virtual environment to support the interaction with a coordinate measuring machine (CMM), which is used to measure dimensions and form of mechanical parts. First, it gives an overview of the implementation, presenting the software "Avalon" and the basic architecture of the developed application. Then, it describes the execution environment, including the description of equipments and devices, characteristics and constraints, and system execution. Finally, it presents the conclusions of the work.

Dynamic alignment system for an FT-NIR Michelson interferometer

Abstract: We present a description of a system for dynamic alignment of a novel Fourier transform Michelson interferometer which is operating in the near infrared (NIR) band. The technique relies on detection of the phase difference between four photodetectors mounted around the main detector. The system is controlled by a computer in real time. It uses a position observer to record the mirror position and with a feedback loop corrects the tilt of the moving mirror in the interferometer. This relatively simple system is very effective and can be applied to other scanning optical interferometers.

An optimized computational technique for free space localization in 3-D virtual representations of complex environments

Abstract: Modeling 3D objects as octrees has demonstrated considerable advantages which led to numerous applications in robotics where free space localization is critical. Efficient neighbor finding techniques in tree structures are required for such models to be used properly, especially for path planning and collision avoidance. In this paper, an optimized neighbor finding approach is presented that is based on a recursive addressing scheme which precludes any backtracking into the tree structure while preserving model compactness. Neighboring cell addresses are computed directly given a displacement direction in 3-D space and the address of the starting cell. Neighboring rule sets that have been previously derived for a quadtree representation are now extended to octrees. Given the algebraic rules that are defined, computation of a neighboring cell address in an octree comes down to basic arithmetic operations with carry. The algorithm complexity is kept low in order to provide good performances.

Music indexing and retrieval

Abstract: Music is a particular case of audio media that has peculiar requirements for retrieval. Its representation, parallelism and multiple features are some examples of the challenges encountered. In this paper, these challenges are characterized and the techniques commonly used for classification, indexing and searching of music content are described. Whenever possible, comparisons are drawn with text information retrieval.

Development of anti-lock brake system in virtual environment

Abstract: With increase of speed, the vehicle's brake performance is of critical importance to its safe and reliable operation. Conventional techniques used to the design and manufacturing of anti-lock brake system (ABS), are based on real analysis and design. While these techniques and control algorithms are time-consuming, and are not always compatible with other kinds of vehicles. This paper presents a new method of design and prototype realization of Vehicle Dynamics Control System-ABS based on virtual environment, which is a generic solution to all kinds of vehicles in real design and manufacturing.

Switched reluctance motor commutation using its magnetisation characteristics

Abstract: Sensorless rotor position detection of switched reluctance (SR) motor has attracted a lot attention in an effort to overcome problems associated with and limitations imposed by the use of a physical position sensor for its control. Most detection schemes advocated so far utilise magnetisation characteristics of the motor of either the normal excitation or that introduced for the purpose of position detection. The effort being directed at achieving varying degree of accuracy and robustness to suit desired application economically while reliability remaining core to every method. This paper present a method developed to control commutation of SR motors using the inductance profile of the stator windings. The method is based on 8-bit processor in an effort to minimise the cost of the detector. The developed scheme was tested successfully online. The paper concentrates on effective and reliable ways of generating commutation using diagnostic signals.

Interaction of free viewpoint 3D video with a haptic-based environment for training applications

Abstract: Traditional 2D video sequences have been characterized as a static and rigid set of scenes. This mature technology faces two major drawbacks: 2D video scenes as a set of standstill and inflexible sequences, and the limited data available to the viewer, that is, no free selection of viewpoint. A novel image media channel has been proposed to record dynamic visual scenario in the real world defined as 3D video. Creativity and imagination has made possible a closer interaction between humans and computers. This interaction depends on haptics, that is, the sense of touch given by computer technology. Virtual reality environment is evolving towards more realistic graphical representations. In this paper we will present a novel approach to integrating free viewpoint 3D video with haptic devices in an environment suitable for training.

A PC-based apparatus for identifying magnetization characteristics of switched reluctance machines

Abstract: This paper presents a flexible-and-friendly apparatus to accurately identify the magnetization characteristics of the switched reluctance machine (SRM) as functions of winding current and rotor position. The identifying method is proposed based on the fully automated measuring setup with a graphical user interface to determine characterization model of the SRM, and take the magnetic nonlinearity and static losses into account. Theoretical basis of the characterization parameters derivation is explored and used to implement the measurement apparatus. Experimental results on a 1-kW, four-phase, 8/6-pole SRM verify the validness of the evolved solution and show very great degree of accuracy. The derivations of the sensitivity and accuracy of this PC-based characterization system are given, and the uncertainty analysis of the system is included. Special emphases are placed on the flexible characterization procedure, computational intelligence and user friendliness scheme of the automatic measurement as well as the capability of the graphical presentation and good documentation of results. The developed instrument not only can be employed to characterize the magnetic circuit of the SRM and linear SRM, but also can be applied to measure the magnetization parameters of the other electromagnetic components or devices.

A development of intelligent and compact active router system

Abstract: Technologies on the active network are a definition of the active packet, the active sensor language, a composition of the active node, and setting a path of the active packets. However, there are few researches about to handle a program code of the active packet in the active router and to control the active router. In this paper, we propose intelligent, controllable, and compact active router system using the LINUX. And we suggest the active sensor language and interpreter executed in the active router system. The active router system is consisted of a packet classifier, a routing process, and applications to execute a program code of the active packet. This router system has a controlling module to install and management programs needed from the active packet's codes.

Modeling moment of inertia for virtual interactive environments

Abstract: Virtual environments have begun using algorithms to model the world around us more accurately by accounting for the mechanical behaviors and properties of objects. Amongst the properties for modeling objects which are used are the moment of inertia, the mass and the center of mass. Many simulators make use of a scalar value for the moment of inertia even though this technique does not accurately represent the anisotropic properties of angular inertia. This paper presents a technique for using an inertia tensor matrix when calculating angular accelerations in virtual environments.

An advanced Java mobile browser using a box model to display the contents

Abstract: The wireless Internet market which uses a wireless terminal is expanding more and more. In addition, development of programs that are to be built in wireless terminal is also growing established mobile browsers, ME and WAP browser, have two modes, browsing mode and Java mode. Two modes do not execute at the same time. This paper proposes method that two modes discharge at the same time in mobile browser using kilobyte virtual machine (KVM) for mobile terminal. We use the box model the renders a text and Java applet on the screen. As this Java mobile browser uses Java, it can download necessary Java applet program on wireless Internet, and it can also execute those applet programs in a terminal.