Tokyo). The idea here is to divide the volume into a set of tetrahedral cells. The amount of calculation is reduced by using the exterior faces of cells rather than the ray as the base for processing. The use of Ray Casting Technique in the rendering of density clouds and surfaces is described by Handana and Takada (both of NEC, Japan). The paper by Tanaka (ATR, Japan) on the visualization of Complex shapes from sparse and incomplete range data is quite impressive. The technique is based on differential geometry using adaptive and arbitrary oriented meshes. Rendering of outdoor scenes by Professor Nakamae reports a very impressive effort at Hiroshima University; they have been constant at work in this problem for a number of years. Their results are maturing very nicely. A group from different Japanese Universities presented a method to render curved surfaces under illumination of parallel cylindrical light sources and to compute the shadows they cast under these conditions. The work is successful in producing quite realistic images. The highlight of the virtual reality session was the paper by Takemura and Kishino (ATR, Japan). Their paper describes a case study of building, a prototype of cooperative work environment. An implementation of an environment where two operators can grasp and move by hand an image is presented. The System uses a hand gesture input system. Three eye-catching presentations were made in the simulated nature session. Modeling of compound leaves from the University of Calgary, simulating bird's eye views from Carnegie Mellon, and synthetic fire works from Japan. Talking about the subject of Computer Animation, this could be accomplished on the basis of kinematics only. This approach often produces unrealistic results and always requires the animator to be heavily involved. More realistic results can be obtained by considering the physical laws and the dynamics of the motion. The obstacle in this approach is the complexity of the models and the computational cost associated with it. At this conference, a number of papers addressed the problem differently, thus overcoming the obstacle. Chua (University of Singapore) presented an adaptive time-step solution scheme for the equation of motion, for rigid and elastic bodies that include a treatment for avoiding collision. The paper, as expected, is mathematical in nature, but the presented animation case of free falling elastic sheet is very interesting. The paper by Amakawa (University of California) addressed the animation of multijoint arm through training a neural network. Two invited papers were also presented on the same subject of computer animation. Each paper had been authored by one member of a couple, Nadia and Daniel Thalmann of the University of Geneva. Daniel presented the animation of growing objects by using L-system. He simulated the interaction with the environment (e.g., existence of wind) through the use of a vector force field. Nadia's talk concentrated on the simplification of parts of the objects through the modification of polygonal meshes. Her objective was to reduce the complexity of the animation of the human body motion and the collision of clothes with it during motion. From the world of animation, the conference turned to the subject of generating 3-D objects by sweeping 2-D ones. A very involved paper by three Korean authors discussed an algorithm to compute an approximation to the general sweep boundary of a 2-D moving object that changes its shape while traversing the trajectory. The method depends mainly on polygonal approximation of the trajectory and the object shape at various points along the trajectory. Each two consecutive instants are linearly connected, then polygons are constructed to approximate the sweep boundary. Another paper by professor Kunii of Japan with two young researchers from Singapore was also presented; Kunii was the Japanese organizer of this conference. The paper describes the integration of the sweep technique with the Homotopy model through the use of one parameter that controls the sweeping as well as the deformation of the cross section shape. The results are quite good. The sessions on "Model Based Coding" contained five papers, all from Japan. Four of the five papers, related to description, Synthesis and Control of facial expressions. In one paper a description is given of a method for the creation of a model and synthesis of facial expressions by using isodensity maps; the technique seems quite successful. Another paper describes the use of input text to control the motion, expression, and mouth shape. The synthetic actors are capable of talking with synchronized correct mouth shape of the sound. The demonstrated results are interesting and impressive. Another paper described a hardware system and software algorithm for visual interactive control of synthesized facial images using finger signs. This required a substantial amount of work by two young researchers from the University of Tokyo. Face-to-Face communication between the operator and the synthetic actor was the subject of a paper by two researchers from Tokyo University and Seiki University. The paper presents a prototype of multimedia man-machine interface based on the media conversion scheme. At the hidden surface and hidden curve session, a good paper by three Japanese professors from three different Japanese Universities was presented. Their algorithm used Bezier clipping, rather than the common polygonal approximation, for the computation of hidden surfaces, interaction of curves, calculation of penetration points, and the extraction of contour lines. In the above, I didn't try to describe all the papers that were presented. My aim was rather to highlight the conference's important papers and those that reflect today's interests, and also to report on major activities. David K. Kahaner, ONRASIA *** WORKSHOP ON MODERN This workshop was held one week after the large Computer Graphics International meeting, held 22-26 June 1992, see my report ["cgi.92", 29 Aug 1992]. The workshop represents the theoretical end of the visualization research area-papers were primarily devoted to combinatoric, topological, singularities, and related aspects of visualization and modelling. A Proceedings containing all the papers was carefully prepared and has been published as follows, therefore I will not review this material in detail but only make a few summary comments. Proceedings: Modern Geometric Computing for Visualization T.L.Kunii & Y.Shinagawa (Eds.) ISDN 0-387-70105-2, Springer-Verlag New York, 1992. Computer visualization requires the computation of various displayable shapes, which are becoming more and more complex along with the objects and phenomena that scientists are trying to visualize. Computation must be fast, and this requires information locality. Such requires information locality. Such information locality is achieved by characterizing the shapes via geometry and topology, and this may need substantial computation that requires supercomputers, but the fundamental supercomputers, but the fundamental tool remains reasoning about geometry. Thus it was not surprising that the most impressive papers were about theoretical tools such as Fomenko's (Moscow) paper on topological classification of Hamiltonian systems, Kergosien's (Paris) on shape modelling through singularity theory, Patrikalakis' (MIT) also on singularity theory for engineering design, or Kunii's (Tokyo) on Homotopy models for a geographical database system. Earnshaw (Leeds) described current scientific visualization work in the UK, actually a very good survey of existing tools and trends. But the high level of other research described at this workshop was emphasized by T.Kunii, who commented to Earnshaw that it was crucial to have a scientific core for any group of research activity. Readers should note that Patrikalakis' work was supported by the Office of Naval Research Research (ONR). I would like to call attention to several papers. Kergosien's work is a careful and complete introduction to Morse theory, Reeb graphs, singularity theory and related algebraic topology as it applies to visualization. The paper by Mori from Ricoh on shape description of characters based on their extremal points. The application here is a faster scheme for optical character recognition. Mori's technique contrasts to that based on Morse theory, and what is impressive here is both the nuts and bolts experimentation that the authors have done as well as the high level of scholarly work from scientists at a company that is usually not highly thought of by the West. (Recently, Mori has also written a survey to be published this year by IEEE: Mori S., Suen CY, Yamamoto K., "Historical Review of OCR Research and Development," IEEE Proceedings, 1992 (in press). Fomenko's paper paper contained some of the deepest analysis, including new results in computer geometry, Hamiltonian mechanics and symplectic topology (in particular studies of the enumeration, recognition, and algorithmic classification of all integrable Hamiltonian systems). Two papers were presented by Melnikov on the development of supercomputers in the former Soviet Union, including a 4 cpu system with 500 MFLOP peak performance. Melnikov is the Director, Institute for Cybernetics Problems, Russian Academy of Sciences. He also described some of the Institute's more current research, including a PCbased handwriting recognition system. Melnikov spoke in Russian, using overhead transparencies of marginal quality. A translator was present, but it was slow going. Fortunately both papers are included, in fairly good English, in the Proceedings, and these represent an important historical record. Although these papers were included in a section titled, "Supercomputing for Modern Geometry," the link to this workshop was very tenuous. Nevertheless, it was a fascinating presentation by someone who was in at the very beginning of the USSR's high performance computing program. The workshop chairperson was Prof Tosiyasu L. Kunii Department of Information Science Faculty of Science The University of Tokyo U-TOKYO.AC.JP FAX: +81-3-3818-4607 who was also the organizer of CGI'92. Kunii is very well known in the West. Along with his students and colleagues, three coauthored papers of his were presented within the application session, including two related to the construction of geographical databases. Many researchers have been interested in terrain visualizations systems for over a decade, but because of the tremendous amount of data, efficient representations and data structures are required. One approach is to somehow employ the concept of a fractal, by determining the fractal dimension of the given terrain data. A disadvantage of this method is that no topological characteristics of the terrain are included, and hence the geometrical characteristics of the terrain are difficult to capture. One research direction of Kunii is to use the homotopy model to describe the terrain in the form of a series of contours and, with certain simplifying assumptions, makes use of the Reeb graph of a 3-D object. In a related paper some guidelines are proposed for the development of a global (geomorphological) database with a 250-m mesh. Finally, a third paper presents a method for reconstructing surfaces from cross-sectional contours based on surface coding theory, also related to Morse theory. These papers involve complicated concepts and difficult mathematics, but have significant potential applications. Program, Modern Geometric Computing for Visualization Tokyo, Japan, June 29-30, 1992 Kogakuin University SESSION 1: Modern Geometric Classification and Its Computation, Computer Geometry and Topological Classification of Integrable Hamiltonian Differential Equations: Visualization of Concrete Physical Examples (invited paper) A. Fomenko (Dept. of Geometry and Topology, Faculty of Math & Mech, Moscow State Univ, Moscow, 119899, Russia. Email: ANATOLY@FOMENKO. MIAN.SU) The Conjugate Classification of the Kernel Form of the Hexagonal Grid, Z. J. Zheng and A. J. Maeder (Victorian Center for Image Processing and Graphics, Dept of Computer Science, Monash Univ, Clayton Vic. 3168, Australia) Shape Description and Classifi cation Based on Extremal Points and Their Relations, Y. Nakajima, S. Mori and H. Nishida (612RG AI Tech Dept., Ricoh R&D Center, 16-1 Shin-ei-cho Kohoku-ku, Yokohama, Kanagawa 223 Japan) Automatic Parallelization of Programs for MIMD Computers (invited paper), V.A. Melnikov, B.M. Shabanov, P. N. Telegin and A.P. Chernjaev (Institute for Problems of Cybernetics, Russian Academy of Sciences, Vavilova St. 37, Moscow, Russia) also (Russian Academy of Sciences, 9, (Russian Academy of Sciences, 9, SESSION 2: Applications of Modern Geometry, Computation of Singularities for Engineering Design (invited paper), N.M. Patrikalakis, T.Maekawa, E.C. Sherbrooke and J.Zhou (Dept of Ocean Engineering, MIT, Room 5-428, 77 Mass Ave, Cambridge, MA 02139-9910 USA, Email: NMP@DESLAB.MIT.EDU) A Geographical Database System Based on the Homotopy Model, T. Ikeda, T. L. Kunii, Y. Shinagawa, and M.Ueda (See above for Kunii's address) A Case Study for Establishing Better Geomorphological Database, M. Ueda, T.Ikeda, T. L. Kunii and Y.Shinagawa (See above for Kunii's address) The Development of the Supercomputer System Electronica SSBIS (invited paper), V.A. Melnikov, Y.I. Mitropolski (see above for Melnikov's address) Closing Remarks: T. L. Kunii, Program Chairperson Conference Organization Computer Graphics Society, The University of Tokyo in cooperation with Information Processing Society of Japan, The Institute of Electronics, Information and Communication Engineers Supported by: Nihon Silicon Graphics Foundation for the Promotion Kubota Computer Inc. - David K. Kahaner, ONRASIA SCIENCE AND TECHNOLOGY AGENCY (STA) SURVEY ON JAPANESE MANUFACTURING RESEARCH AND DEVELOPMENT, 2 JULY 1992 The Japanese STA recently concluded an investigative report on the research activities of Japan's private sector manufacturing firms. In compiling this report, the STA surveyed 1,301 manufacturers with a level of capitalization over 1-billion yen, eliciting a 63.9% response. According to the results of the investigation, R&D expenditures of the respondent companies totalled 7,402.20-billion yen against sales of 207,700-billion yen in 1990. The average level of R&D expenditure per manufacturer was 8.9-billion yen. By industry, average levels of expenditure in billions of yen was as follows: measuring instruments industries at 8.6%. The ratio for the precision machine industry and the electric machine tool industry stood at 6.7% and 6.5% respectively. STA used the responses to estimate the actual number of researchers working for manufacturing companies. The number was 215,000. The communications, electronics and electronic measuring instruments sector led in this category with 701 researchers per company. This same sector also led in percentage of employee involvement with 13.7% of payroll expenditures involved in R&D. This figure was more than double the average for all industries, which stood at 6.2%. Information submitted to STA over the course of this survey indicates that Japanese manufacturing companies held approximately 369,000 domestic patents. A breakdown of this figure by industry shows that an average of 1,612 domestic patents are held by each company in the communications, electronics and electronic measuring instruments sector. As a function of capitalization, an average of 92 patents are held by manufacturing companies capitalized at between 1- and 5-billion yen, and an average of 3,499 patents are held by manufacturing companies capitalized at over 100billion yen. According to STA, average R&D expenditures at overseas subsidiaries where Japanese manufacturing companies have over a 50% ownership stake stands at 367.9-billion yen. This figure is equal to approximately 1.28% of sales for these companies. The automotive sector spent most on R&D overseas, and it was followed by the electric machinery sector. This survey also included a series of questions designed to illustrate some trends in international R&D activity by the Japanese manufacturing companies. The survey revealed that a group of 117 Japanese manufacturing companies have a total of 276 R&D facilities located overseas. A breakdown of this number by industry shows that 19 manufacturing companies in the electric machinery sector have a total of 54 facilities overseas. The precision machinery sector ranked second with 13 manufacturing companies that have a total of 36 facilities. While the majority of these facilities have been located within the United States, future planning on the part of Japanese manufacturing companies indicates a trend towards locating new R&D facilities within Europe and the Asian countries. The necessity of improving products to meet consumer needs was the most frequent justification given for the decision to locate R&D facilities over seas. This survey uncovered some interesting attitudes within Japanese manufacturing companies towards technology transfer. According to survey responses, 36% of Japanese manufacturing companies expected that "technology can be expected to be both exported and imported." On the other hand, 13% of the companies responding expressed the expectation that technology "can be expected only to be exported," 15% stated that "it will only be imported,"and 36% stated that "no technology transfer will occur." Regarding the destination to which technology would be exported, 43% of Japanese manufacturing companies stated that it would be exported "only to companies other than our subsidiaries," while 27% chose the response "more to companies other than our subsidiaries." As part of this survey, participating Japanese manufacturing companies were asked to rank the different regions of the world in terms of comparative R&D capabilities over time. Results of these questions indicate that at present, most Japanese companies still perceive the United States to have the strongest R&D capabilities, Europe follows, and Japan is in third place. But, five years in the future, Japan will supersede both the United States and Europe. Finally, the survey indicated that most Japanese manufacturing companies expect competition among the United States, Japan, and Europe to continue and tension to increase. In an attempt to identify countermeasures to this expected increase in tension, 36% of Japanese manufacturing companies suggested cooperation in research and development, 28% suggested technology transfer, and 12% suggested the establishment of overseas research facilities. Daivd K. Kahaner, ONRASIA OKI ELECTRIC'S SOFTWARE IMPROVEMENT ACTIVITIES In a recent report, ["oki.ps", 9 May 1992], I described some of the software production research that was going on at Oki Electric. The techniques that large Japanese software producers use to address problems of creating, managing, and maintaining software are clearly of interest to others. A recent book, Japan's Software Factories, Michael A. Cusumano, Oxford University Press, 1991 contains many anectodal descriptions of how this is done in Japan. Scientific readers of my reports should note that the largest software products, containing millions, or even tens of millions of lines of source program text, are often found outside the technical arena, in securities, insurance, and other financial industries. Communication and command and control software can also be massive. Recently, I was accompanied by Mr. Allan Willey Email: willey@mot.com on a visit to Oki to hear about their recent efforts firsthand. (Oki is not one of the companies that Cusumano details in his book.) Obviously, Motorola is interested in software management, as very large fractions of the value added aspects of their cellular telephone infrastructure are due to software. Mr. Willey commented that companies are looking for ways to reduce the dependence on software stars, because of the high risks that individual organizations incur when people leave, become ill, etc.. These companies are also trying to verify and understand the Japanese software developers claims of 10% productivity improvements annually. I wish to thank Mr. Willey for his summary of his visit, which I have merged into the report below. For those people not familiar with Oki, this was Japan's first private telecommunications manufacturer that produced a prototype telephone in 1881, barely five years after Bell's invention. Last year Oki's sales amounted to 661 Billion Yen, close to US$ 5 Billion, and has slightly more than 20,000 employees. Oki has three main business activities, |