Image retrieval, Categorization, Internet, World Wide Web In recent years, the WWW (World Wide Web) has grown up explosively and image information has been widely distributed. However, existing information retrieval systems on the WWW can hardly deal with image data. We have been developing a new type of information retrieval system for image data on the WWW. In this paper, we discuss problems of image retrieval and filtering on the WWW. Then we propose the open image retrieval system GIRLS (Global Image Retrieval and Linking System) that enables flexible retrieval and filtering according to a user's view. We show examples of retrieval.

View full abstract

This paper describes an information retrieval system which offers real timeinformation from mobile vehicle through Internet. More precisely, in order to test the efficiency of solar battery and electric vahicle, four crew members of Tokyo Denki University made a round trip through North American Continent by their hand made solar electric vehicle this summer. In accordance with this project, an information retrieval system which offers the daily drive information through internet to the public was constructed. The system configuration and operation results of this system are reported.

View full abstract

We developed an experimental system for appreciating solid object fine arts like a china or a sculpture object. The features of the system are that you can appreciate it just as if you were handling the fine art with your hands, the feature of each fine arts has been shown intelligibly using image, sound and text, the high quality image of fine-arts can be displayed as real-time, image of fine arts can be stored up without losing the impression of actual objects, etc. We confirmed the validity and the good operability for appreciating solid objects fine-arts, using the experimental system which consists of the Hi-VISION video server in which random access of real-time is possible in 1/60 seconds, manipulator for the system, fine-arts capturing system, etc. In the future, such an appreciation system will be expected to be used effectively in the museum, the educational organization, etc.

View full abstract

In this paper we are proposing a file system with compression and decompression processing capabilities for a teletext receiver. By improving memory management and eliminating various memory capacity problems, we have created a system with minimal memory requirements. The file system encompasses a teletext data format, a protocol for its transmission and an application for the compression and decompression of the data. Furthermore, the file system can also be used under a multi-tasking OS because it can handle simultaneous file accesses.

View full abstract

In recent years, many methods have been proposed for presuming 3-D shapes from features in a 2-D image. Most of these methods are carried out under the assumption that some of the 3-D coordinates of the object are known. This study proposes a method for recognizing the 3-D shape of a target object without any 3-D coordinates of the object. The method utilizes two kinds of informations : one is the shape of a single cross section of the target object, and the other is the sequence of several 2-D images of the rotating target object. By deforming a 3-D wireframe model based on these informations, the shape of the 3-D object can be obtained.

View full abstract

We propose an image restoration technique through vector quiantization and codebook mapping. To reduce quantize distortion, the idea of fuzzy vector quantization is introduced. Further, fuzzy mapping and fuzzy histogram approximation algorithms are used to improve mapping accuracy. Finally, the usefulness of the proposal method is examined by a simulation.

View full abstract

In order to find object representation method which can unify local information, such as tangent vector fields or normal vector fields, and global recognition. The authors have proposed a method which is acle to recognize objects invariantly under Euclidean motion, using linear Lie subalgebra of perceptional vector field of their contours. This subalgebra can represent a very wide class of non algebraic shapes, although they are often non-compact. This paper presents a novel method to represent objects using Lie algebra of Hamiltonian vector field which are has stronger descriptive power and can represent compact shapes. We also show the complete set of invariants of objects under Euclid motion.

View full abstract

Currently, simplex models are mainly used for Representation model of 3D objects. Since this model is hard to build and no natural the pyramid structure associated with it, we propose in the paper a new model based on cubical singular simplicial complex. We also extend the K-map and show that the cubical singular simplicial complex model can be easily acquired by the extended K-map then. Then, we show procedure to build pyramid for multi resolution model. Fast algorithms to calculate topological invariant of such as homology group and homotopy, are presented.

View full abstract

This paper reports education and research of computer science at University of the Ryukyus, especially at Department of Information Engineering. The main feature of the education system are : (1) We adopt Core Curriculum and assign a few required subjects to give students much freedom to select subjects.(2) The freshmen are assigned a training course to survey computer science in a top-down manner.(3) In this course, the students master the computer-literacy as computer users, and the literacy is effectively used in an advanced course for researchers and developers of computer. In order to realize this, we provide a lap-top computer to each student, and let students use the computers to read and write a network news and an electronic mail in their class room and student laboratory and also in their daily life. Some faculties of the university study on not only high tech of computer but also local theme of Okinawa.

View full abstract