|
|
 |
|
|
Multimedia Systems Laboratory |
Noriaki Asada
Professor |
William L. Martens
Associate Professor |
Yasushi Kikuchi
Assistant Professor |
Jung-pil Shin
Assistant Professor |
|
Recognition, analysis and expression of images and sound are studied in the multimedia systems laboratory.
The 3-D motionand the 3-D shapes of the object are analyzed, understoodand expressed bycomputerfromthe images taken bymultiTVcamera or video camera in the study of computer vision and/or the 3-D stereovision. Detailed movies can be seen from any angle through the 3-D video system, which is expected to be applied to many fields such as the sports coaching and the recognition of the human facial expression.
In the remote sensing, the category classification technique can analyze the land usages from color information of land images. A laboratory member is engaged in theSELENEproject, the Japanese lunar explorer mission, to analyze the lunar mineral composition and to investigate the future moon surface land usage. Moreover, by applying this technique to the computer vision, even the living body information of the human being can be known, too. In the on-line handwriting character recognition system, the fundamental study to use the handwriting letter as the input system of the computer is investigated. The character recognition is one of the most important fields on the image processing.
The visualization technique is one way to improve the understandings of the mathematics problems. For example, when examining a three-dimensional graph with complicated function, a four-dimensional graph can be easily recognized, where time, color, sound and so on is introduced as the fourth dimensions. Also, it is very useful to understand the mathematical structure of the function to display zero points and limitation values visibly. Just as3D graphic rendering can produce realistic views of virtual spaces, so can3D audio rendering produce realistic sounds for objects and events in those virtual spaces, called "Virtual Acoustic Space". State-of-the-art 3D audio rendering is based upon acoustical measurements. Laboratory members have participated actively in scienti??c meetings, both in Japan and abroad. They have presented and participated in seminars, and in presentation of scientific results in fully refereed publications.
The laboratory is the coordinator for course-work in astronomical observation by cooled CCD cameras. Students learn how to use equipment to get some data into computer system, how to process image data, how to analyze these data and how to study the nature through this course-ware project.
|
| [jpshin-001:2001] | Jintae Lee and Jungpil Shin. Modeling Virtual Papers for Dynamic Simulation of Ink Diffusion. Journal of Three Dimensional Images, 15(1):155-159, 2001.
A technique is presented for modeling virtual paper model that enables dynamic simulation of ink diffusion on an interactive digital painting system. With the paper model, users can represent wide range of randomly generated fiber meshes, providing a basis for dynamic ink di??usion. The image diffusion algorithm is e??ective for creating diffusion images produced on different types of paper. The resulting diffusion images reflect the texture and the global change of gray tone as well as the local temporal variations specific to the paper. Examples that demonstrates the capability of the proposed technique are presented for both the liquid ink and paper types. |
| [jpshin-002:2001] | Jung pil Shin. A Study on Stroke-to-Stroke Correspondence for On-Line Character Recognition. Journal of the Korean Institute of Industry and Technology, 3(1):32-39, 2001.
This paper describes annew stroke-to-stroke correspondence search algorithm for application to stroke-order-free online character recognition. The hard problem of stroke-to-strokecorrespondence search is simplify from factorial to exponential orderalgorithmcomplexity, which is achievedby using cubic form automaton whose nodes correspond to matching status of reference strokes to input strokes.The optimal path search for stroke correspondenceis performed by Dynamic Programming. The efficient search method by incorprating beam search technique is also investigated. By experiments, it is demonstrated that the proposed algorithm works fast enough for practical application. |
| [jpshin-003:2001] | Jung pil Shin. Online Handwriting Character Analysis Using Stroke Correspondence Search. Journal of Shanghai University, 5:139-143, 2001.
In this paper, the useful information about stroke-orderandstroke-connection are investigated for the stroke order- and number-free online handwriting recognition. Using an automatic stroke correspondence algorithm, we perform the style analysis of stroke-order variation and connection between strokes from a set of the handwritten Chinese characters. The majority of real characters were written in fixed stroke-order, and stroke-order is predominantly in the standard stroke-order. Almost all stroke-connectionsoccur in the standard order, whereas 2 stroke-connections occurred often, and stroke-connections in nonstandard order occurred very rarely. In a comparison of the expected stroke-connections, very few connections were found to actually occur. Furthermore, the methods for incorporating the information are described on the completely stroke order- and number-free framework in order to improve the computational time and recognition accuracy. |
| [jpshin-004:2001] | Jung pil Shin. A Study on Stroke Distance for Stroke Order Free On-Line Handwriting Character Recognition. Journal of Three Dimensional Images, 15(4):146-151, 2001.
For on-line handwriting character recognition, this paper is a comparative study of evaluation methods for individual stroke information on the shape and position of each stroke and relative information on the mutual relation among strokes. First, in the stroke-order free framework, the evaluation of the individual stroke information is performed by introducing the previously proposed measures and new measures based on dynamic programming. Second, the additional evaluation of the relative information is performed in the framework using the optimal path search for stroke correspondence that includes both individual and relative information on stroke. The relative information using the vectors between ending points is introduced.Experiments show that recognition accuracy is very e??ectively improved by applying dynamic programming method to the evaluation of individual information, and prominently improved by additional use of relative information. |
| [wlm-001:2001] | A.Marui and W.L. Martens. Multidimensional Perceptual Scaling of Tone Color Variation in Three Modeled Guitar Ampli??ers I: Deriving Stimulus Space Using INDSCAL. Journal of Shanghai University (English Edition), 5(Suppl.):229-234, 2001.
Multidimensional perceptual scaling analyses were performed for a set of prerecorded guitar performances processed using models of three characteristic types of amplifier distortion (British Crunch, Combo 335, and Twin Drive), and the tone color of the output was varied using three nominal output character settings (termed Normal, Edge, and Punch). This analysis separated out two salient tone-coloration dimensions from a third dimension of that was related to variation in amplifier distortion. |
| [wlm-002:2001] | W. L. Martens and A. Marui. Multidimensional Perceptual Scaling of Tone Color Variation in Three Modeled Guitar Amplifiers II: Interpreting Stimulus Space via Adjective Rating. Journal of Shanghai University (English Edition), 5(Suppl.):235-239, 2001.
The multidimensional perceptual scaling analyses performed in a previous study revealed the underlying dimensions describing variation in amplifier distortion. In this continuation of the previous study, the interpretation of those dimensions was enabled via the results of a semantic differential scaling experiment. |
| [wlm-003:2001] | M. Cohen and W. L. Martens. Spatial Media Research at the University of Aizu. JVRSJ: J. of the Virtual Reality Society of Japan, 6(2):140-145, 2001.
This review concentrated on aspects of multimedia research of the Spatial MediaGroup "at the University of Aizu. Includedwere outlines of curriculum, facilities, research projects, and annual " Spatial Media seminar series. |
| [wlm-004:2001] | S. Lambacher, W. L. Martens, B. Nelson, and J. Berman. Identification of English voiceless fricatives by Japanese listeners: The influence of vowel context on sensitivity and response bias. Acoustical Science and Technology, 22(5):334-343, 2001.
In order to examine the ability of native Japanese speakers to distinguish between English voiceless fricatives, a five-alternative, forced-choice (5AFC) test was administered to104Japanesestudents of English. The results showed that both vowel environment and consonant context had an eect on the listeners' perception of the stimuli in this study. |
| [wlm-005:2001] | W. L. Martens and M. Cohen. Virtual Acoustic Research at the University of Aizu. JVRSJ: J. of the Virtual Reality Society of Japan, 6(3):261-269, 2001.
This review concentrated on the virtual acoustic research conducted within the "Cyberspatial Audio Group" at the University of Aizu. Included were descriptions of available facilities, overview of the problem areas of the field, and a summary of the research results of student projects. |
| [wlm-006:2001] | W. L. Martens. Display of nearby virtual sound sources outside the user's field of view: The dark voice channel. The Journal of Three Dimensional Images, 15(4):87-95, 2001.
A prototype bimodal display was implemented that separates the display space into two distinct regions one for presentation of binocular visual information, the other for presentation of binaural auditory information. The user of a wide-angle, stereographic visual display receives a personal presentation of virtual sound sources very near the user's head, but from behind, outside of the user's field of view. |
| Referred Proceeding Papers |
| [jpshin-005:2001] | Jungpil Shin. Stroke Correspondence Search for On-Line Character Recognition Using Intra- and Inter-Stroke Information. In Editor H.R.Arabinia, editor, International Conference on Artificial Intelligence, pages 272-278, Las Vegas, June 2001. CSREA Press.
As a means to perform stroke order-free and stroke number-free on-line character recognition, this paper describes a new systematic stroke correspondence search algorithm. Inter-stroke information on the mutual relation among strokes and conventional individual intra-stroke information on the shape and position of each stroke are both employed. This search algorithm simultaneously performs both (i) stroke correspondence and (ii) stroke connection, further using both (iii) inter-stroke information and (iv) intra-stroke information. Experiments demonstrate high recognition capability, as well as computational feasibility. |
| [jpshin-006:2001] | Jungpil Shin. Online Cursive Hangul Character Recognition Based on Dynamic Programming. In Bob Werner, editor, Sixth International Conference on Document Analysis and Recognition, pages 810-814, Seattle, Washington, Sep. 2001. IEEE Computer Society.
In this paper, a new algorithm for the on-line recognition of cursive Korean characters (hangul) is investigated for the problem of searching the segmentation points to divide the phonemes. The segmentation points searching is systematically performed by two-level DP matching algorithm, that is, one algorithm targets the phoneme unit level, and the other the whole character unit level. Basedon thecomposition characteristics of hanguls, effective syntax is introduced as part of the character level proces. Fine discrimination is effectively realized for both phonemesand characters including the individual information of each stroke and mutual information among strokes. Experiments demonstrate real time feasibility, high recognition and segmentation performance. |
| [jpshin-007:2001] | Jungpil Shin. A Study on Stroke Distance for Stroke Order Free On-Line Handwriting Character Recognition. In Proceedings of the Fourth International Conference on Human and Computer, pages 1-6, Japan, Sep. 2001.
For on-line handwriting character recognition, this paper is acomparative study of evaluation methods for individual stroke information on the shape and position of each stroke and relative information on the mutual relation among strokes. First, in the stroke-order free framework, the evaluation of the individual stroke information is performed by introducing the previously proposed measures and new measures based on dynamic programming. Second, the additional evaluation of the relative information is performed in the framework using the optimal path search for stroke correspondence that includes both individual and relative information on stroke. The relative information using the vectors between ending points is introduced. Experiments show that recognition accuracy is very effectively improved by applying dynamic programming method to the evaluation of individual information, and prominently improved by additional use of relative information. |
| [wlm-007:2001] | W. L. Martens. Two-subwoofer reproduction enables increased variation in auditory spatial imagery. In Proc. International Workshop on SpatialMedia, pages 86-97, Aizu-Wakamatsu, Japan, 2001.
Although only a single subwoofer is typically used in two-channel and multichannel stereophonic sound reproduction, the use of two subwoofers enables manipulation of low-frequency interaural cross-correlation(IACC)and this manipulation is particularly effective in producing variation in auditory spatial imagery. In order to document this variation objectively, a series of listening experiments were executed using a set of stimuli generated at five correlation values and presented in two reproduction modes. Both modes used two subwoofers, but in one of the reproduction modes identical signals were applied to the two subwoofers. The results of both exploratory and confirmatory listening experiments showed that the range of variation in both perceived auditory source width (ASW) and perceived auditory source distance (ASD) is reduced when negative lycorrelated signals are not reproduced at low frequencies. Global dissimilarity judgments were made for this set of ten stimuli in an exploratory study designed to reveal the salient perceptual dimensions of the stimuli. A subsequent confirmatory study employed a two-alternative forced-choice task in order to determine how identifiably different the stimuli were with respect to the two perceptual attributes revealed in the exploratory study, those two attributes being ASW and ASD. |
| [wlm-008:2001] | W. L. Martens. Binaural range display with independent control over loudness: Presenting nearby whispers in the dark. In Proc. ICAT: Int. Conf. Arti??cial Reality and Tele-Existence, pages 100-109, Tokyo, 2001.
This paper focuses upon the development of an e??ective binaural display tech-
nology employing a simpli??ed model of range-dependence in simulated head-
related transfer functions forheadphonedisplay of virtual sources at close range
(within the user's personal space). The presentation of whispered speech orig-
inating from within the blind area just behind the user has been termed the
\dark voice channel. User tests con??rmed that the prototype auditory dis-
play supports the useful feature of varying the auditory range of equally loud
nearby whispered speech. Based upon psychophysical functions for auditory
range derived under conditions of expected use of the display, a calibrated de-
ployment of the range control was accomplished. In contrast to most binaural
display technology, then, this prototype system supports independent control
over source loudness and range.
Unrefereed Papers |
| [wlm-009:2001] | K. Honno, W. L. Martens, and M. Cohen. Psychophysically-derived control of source range for the Pioneer Sound Field Controller. In Proc. 110th Conv. of the Audio Engineering Society, Amsterdam, Netherlands, 2001. Preprint 5313 |
| [wlm-010:2001] | A. MaruiandW.L. Martens. Perceptual andSemantic Scaling for User-Centered Control Over Distortion-Based Guitar Effects. In Proc.110th Conv. of the Audio Engineering Society, Amsterdam, Netherlands, 2001. Preprint 5387 |
| [wlm-011:2001] | W. L. Martens. Uses and misuses of psychophysical methods in the evaluation of spatial sound reproduction. In Proc. 110th Conv. of the Audio Engineering Society, Amsterdam, Netherlands, 2001. Preprint 5403 |
| [wlm-012:2001] | A. Marui and W. L. Martens. Individual Differences in User Preference for Distortion-Based Guitar Effects. In Proc. 10th Regional Conv. of the Audio Engineering Society, Tokyo, Japan, 2001.
Preference judgments from a group of twelve potential users of a multi-effects processor were collected for all pairwise comparisons of 15 different guitar-sound stimuli. These pairwise preference data, along with pairwise dissimilarity judgments for the same set of stimuli, were submitted to MULTISCALE software and PREFMAP-3 software performing INdividual Differences SCALing (INDSCAL) and MultiDimensional Unfolding (MDU) analyses, respectively. The INDSCAL analysis produced a spatial configuration of the 15 stimuli in the form of a group Stimulus Space expressing the common dimensions underlying the perception of the stimuli for the entire group. The MDU analysis produced a spatial configuration of ideal stimulus points in that same Stimulus Space for each of 12 listeners. |
| [wlm-013:2001] | W. L. Martens. Psychophysical calibration for controlling the range of a virtual sound source: Multidimensional complexity in spatial auditory display. In Proc. Int. Conf. on Auditory Display, pages 197-207, Espoo, Finland, 2001. ICAD. (Keynote address) |
| [jpshin-008:2001] | Jung pil Shin. Research for On-line Hangul Character Recognition, Japanese Ministry of Education Grant-in-Aid for Encouragement of Young Scientists (A), Contract No.12780290, 2000-2001. |
| [wlm-014:2001] | W. L. Martens, 2001. Active member, AES Technical Committee on Perception and Subjective Evaluation of Audio Signals, Audio Engineering Society (AES) |
| [jpshin-009:2001] | Mayuko Obora:s1061038. Graduation Thesis: Feature Analysis for On-Line Handwriting Recognition, University of Aizu, 2001.
Thesis Advisor: Jungpil Shin. |
| [jpshin-010:2001] | Kazuyuki Kunimori:s1061063. Graduation Thesis: Signature Verification Using Pen Inclination, University of Aizu, 2001.
Thesis Advisor: Jungpil Shin. |
| [jpshin-011:2001] | Hitomi Suzuki:s1061095. Graduation Thesis: On-Line Hand-written Kanji Character Recognition Using Relation to Relative Position of Strokes, University of Aizu, 2001.
Thesis Advisor: Jungpil Shin. |
| [jpshin-012:2001] | Yuji Tsuchida:s1061109. Graduation Thesis: On-line Handwriting Character Recognition Using Generalized Learning Vector Quantization, University of Aizu, 2001.
Thesis Advisor: Jungpil Shin. |
| [wlm-015:2001] | Takahiro Nakamura. Graduation Thesis: Development and Evaluation of Software-Based Distortion Effects for Realtime Audio Input, University of Aizu, 2001.
Thesis Advisor: William L. Martens |
| [wlm-016:2001] | Ryo Ito. Graduation Thesis: Development and Evaluation of Software-Based Delay Effects for Realtime Audio Input, University of Aizu, 2001.
Thesis Advisor: William L. Martens |
| [wlm-017:2001] | Yasuyuki Kato. Graduation Thesis: Development of a Java-Based Graphical User Interface for Guitar Effects Software, University of Aizu, 2001.
Thesis Advisor: William L. Martens |
| [wlm-018:2001] | Masataka Abiko. Graduation Thesis: A Java-Based Graphical User Interface to Online Guitar Effects for Beginners, University of Aizu, 2001.
Thesis Advisor: William L. Martens |
| [wlm-019:2001] | Yoshifumi Tomita. Graduation Thesis: Development of a User-Centered Control Structure for Software-Based E??ects Guitar Effects, University of Aizu, 2001.
Thesis Advisor: William L. Martens |
| [wlm-020:2001] | Kazuhisa Miyashita. Graduation Thesis: Multidimensional Perceptual Scaling Of Tone Color Variation In Three Modeled Guitar Amplifiers, University of Aizu, 2001.
Thesis Advisor: William L. Martens |
| [jpshin-013:2001] | Jung-pil Shin. Presentation of Programs for Handwriting Recognition. Festival of University of Aizu, Oct. 2001. |
| [wlm-021:2001] | C. C.W.Giragama. University of Aizu/ Masters Thesis: Relating multilingual semantic scales to a common timbre space, 2001.
Thesis Advisor: William L. Martens |
|
Annual Review 2001