The sign-language can be used as a communication means between avatars having no common language. As a trial to overcome the linguistic barrier, we have previously developed a 2D model-based sign-language chatting system between Korean and Japanese on the the Internet. In that system, there have been some problems to be solved for natural animation and real-time transmission. In this paper, we employ a 3D character model for stereoscopic gestures in the sign-language animation. We also utilize CG animation techniques which use the variable number of frames and a cubic spline interpolation in order to generate realistic gestures. For real-time communication, on the other hand, we make use of an intelligent communication method on a client-server architecture. We implement a preliminary communication system with Visual C++ 5.0 and Open Inventor on Windows platforms. Experimental results show a possibility that the system could be used for avatar communications between different languages.

This paper describes an experimental system for designing three-dimensional solid objects through three-view orthographic drawings, using automatic three-dimensional interpretation in cooperating with a set of interactive drawing-oriented design utilities. In the design process based on the system approach, an object is viewed as being consisting of several volumetric instances, each of which is further composed a number of simple subvolumes. In the system, whereas the automatic interpretation is employed to construct the instances from their orthographic projections, the design utilities are provided to allow modifications of a solid-object model being composed of the constructed instances. Once a solid-object model is created from a drawing containing the projections of an instance, modifications on the drawing can be performed for addition of new instances into the existing solid-object model, and changes and deletion of the instances existing in the model. Being aimed at a two-dimensional interface for the Constructive Solid Geometry, or CSG method, the system provides the user with design interaction procedures which let the user manipulate the projections of the instances in terms of two-dimensional graphical primitives, such as straight lines, arcs, and circles, obtain and display a solid object constructed from the projections being designed. In addition, more complex three-dimensional object shapes can be achieved using this alternative approach after the interpretation concept.

We present experimental results for a structural learning method of feed-forward neural-network classifiers using Principal Component Analysis (PCA) network and Species Genetic Algorithm (SGA). PCA network is used as a means for reducing the number of input units. SGA, a modified GA, is employed for selecting the proper number of hidden units and optimizing the connection links. Experimental results show that the proposed method is a useful tool for choosing an appropriate architecture for high dimensions.

Existing solid models often contain redundant primitives and null blocks, which both slows down the rendering process and makes the process complex. There has been recent progress toward solving this problem, but existing modeling schemes cannot support eliminating all the redundancies, especially the null blocks, from the solid models. This paper proposed a technique that can eliminate redundancies. By dividing a primitive into some surface dispersed points, a new primitive representation is obtained. The sample segments of the primitive or the object are used to locate composition position to prevent the null primitives from being generated. By drawing out the geometric shape points set corresponding to a common acting area, the volume boundary of a primitive or an object is evaluated by only the Boolean set operations. The null blocks can be picked out in terms of the volume boundary. The resulting solid model generated in this way has no redundancies and is suitable for fast rendering of the image.

Active net is a deformable model which utilizes the network analogy of a physical region. In the model, the region of a target is detected by minimizing the energy defined for the sample points of the model. The region of the target is extracted using fixed network topology in the orginally proposed algorithm. In this paper, we introduce the network reconfiguration mechanisms such as tearing and division to realize multiple objects detection and complex object detecion. The introduced algorithm dynamically unlinks the arcs of the network when their strain value exceeds predefined threshold level. In the method, we propose a new image energy which improves the position sensitivity of edges without increasing computation cost. Experimental results for images taken by video camera show the validity of our approach.

To employ Computer Graphics (CG) technology in the production of CG holograms or similar applications where the generation of consecutive pictures with rotated viewing directions is demanded takes a lot of time and money. This paper presents a two-pass surface rendering method for fast creating such animated pictures from a single view image structure. Instead of generating display data for a single picture, the first pass program constructs a Position Oriented (PO) frame buffer that contains the information of control points on every order of visible surfaces. A display program of second pass manipulates the frame buffer structure to generate surface rendered pictures with spatial and perspective transforms. Experiments proved the method is effective to create high quality surface rendered images with rotation capability. Various usuful facilities for CAD visualization have also been developed within the technique. The improvements on memory consumption and picture quality are also discussed in this paper.

In this paper, we propose a method to construct structure models of articulated objects from multiple local observations of their motion using state transition analysis of local geometric constraints. The object model is constructed by a bottom-up approach with three levels. Each level groups sensor data with a constraint among local features observed by the sensor, and constructs the local model. If the sensor data in current model conflict, the model is reconstructed. In each level, the first level estimates a local geometric feature from the local sensor data (eg. edge, feature point) The second level estimates a rigid body from the local geometric feature. The third level estimates an object from the rigid bodies. In the third level, the constraint between rigid bodies is estimated by transition states, which are motions between rigid bodies. This approach is implemented on a blackboard system.

The use of the Constructive Solid Geometry (CSG) model has been considered in several computer-aided design systems for recent years, since its concept of powerful and multiple operations on basic object shapes to create more complex ones is intuitively easy to understand. Based on the CSG concept and a guiding principle of scene analysis, an algorithm for interpreting three-view drawings is described in this paper. In a certain process, the 3D interpretation works on local recognition to detect possible subparts of which orthographic patterns are defined in the 2D description of a part. In the subsequent process, which examines all possible solutions to the given drawing until finding one that meets some goal criteria, the combinatorial operators are applied to the possible subparts. Through those repeated processes, the final solution can be obtained in terms of consistent subparts in a CSG tree, including additional information about solids and cavities.

In order to investigate the possibility of avatar communication using sign-language, in this paper, we develop a sign-language chatting system on the Internet using CG aniamtion techniques between Korea and Japan. We construct the system in server-client architecture, where images of Korean or Japanese sign-language are analyzed into a series of parameters for sign-language animation by server. We transmit the parameters, which are text data instead of images or their compression, to clients and regenerate the corresponding CG animation using the received data. The chatting system is implemented with Visual C++ 5.0 on Windows platforms. Experimental results show that the sign-language could be used as a communication means between avatars of different languages.

This paper describes the research results of three dimensional parameter representation of human arm motion, image synthesis and intelligent transmission for sign language communication between Japan and China using a personal computer.

Two new algorithms are presented for determining the elements of the matrix function Al of a digital system of which the matrix of dynamics has nonzero elements only in its superdiagonal (ones) and in its last row (a particular combination of the coefficients of the transfer function denominator). Advantages and disadvantages of these algorithms are discussed. It is shown that when using the matrix function Al only for the calculation of the system response, the second algorithm presented here seems to be particulary useful due to the simplicity of its application and due to the fact that it allows us to omit the calculations of the eigenvalues of the matrix of dynamics, which seem to have been necessary for most of the known methods. It also reduces the number of computer operations which must be performed from n3(n-2) using direct matrix multiplication, to n(n-2). The second algorithm can be efficiently used for the determination of the elements of limited power of the matrix of dynamics. Both of the algorithms presented in this paper are particularly useful when applied to the matrix of dynamics of a digital filter, of which the transfer function is presented in the direct form of realization, since, the matrix of dynamics of such a digital system is expressed directly in a proper form.

This paper introduces a calligraphic character writing system in which the generated characters are really written out on paper by a robot holding a writing brush. Being different from the conventional researches, the data base is made by means of a new method in which features of different style of calligraphic characters can be extracted with a uniform algorithm. Besides, we really constructed a robot for writing generated characters and the problems introduced by adopting a brush and their solutions are also discussed. In particular, because of the softness and friction against paper of the brush head, the postponement of the brush touch may be caused and we solved it by developing a special compensation method. In addition, an experimental result based on several style of characters is described. It turns out that, since the writing process is much similar to the one made by man, the aesthetic perception and momentum as well as some intrinsic properties of calligraphy can be better demonstrated.

A two-cascaded image processing approach to enhance the subtle differences in X-ray CT image is proposed. In the method, an asymmetrical non-linear subfilter is introduced to reduce the noise inherent in the image while preserving local edges and directional structural information. Then, a subfilter is used to compress the global dynamic range of the image and emphasize the details in the homogeneous regions by performing a modular transformation on local image den-sities. The modular transformation is based on a dynamically defined contrast fator and the histogram distributions of the image. The local contrast factor is described in accordance with Weber's fraction by a two-layer neighborhood system where the relative variances of the medians for eight directions are computed. This method is suitable for low contrast images with wide dynamic ranges. Experiments on X-ray CT images of the head show the validity of the method.

Voxel representation of solid object is a well known model that is widely used in CAD, robots, computer graphics, and medical image processing. Efficient algorithms for the tracing in voxel model are of great importance. A fast hardware oriented voxel tracing algorithm (called LCDDA) is proposed and implemented. All the voxels pierced by a given ray are generated incrementaly. To meve from a voxel to the next one only three fixed point additions and comparisons, which can be done using 3 parallel pipelines, and a ROM access are required. It takes 0.2 microseconds to do this under 5 MHz clock on a personal computer, which is 103 times faster than Fujimoto's and 1339 times than Glassner's algorithm on a workstation. As an application of this, time consuming ray tracing process is performed by this hardware implemented algorithm on a personal computer. Also, an active intersection table (AIT) is introduced to avoid multiple intersection of a ray and the same object. The high speed of these methods have been proved by experimental results.

Expansive graph grammar is one of the suitable tools for describing pattern structures in syntactic pattern recognition. To describe the numerical features of patterns and relations among them simultaniously with pattern structure, in this paper we show with expansive graph grammar: (1) how to describe the dependent relations among attributes of (sub) patterns with the concepts of inherited and synthesized attributes, (2) how to describe conditions supposed to be satisfied by attributes of patterns with the concept of predicate. An interpreter for such processing is presented based on the Problem-Reduction Representation (PRR). Some features of the method and the interpreter are discussed. It is showed that the method introduced in this paper is suitable for describing patterns in scene analysis when numerical features of patterns are considered.

It is an important topic to express solid objects in CAD, CAM and CG with a uniform way. In this article, a constructive dispersed surface points-set model (CDSPM) is proposed. CDSPM is made up by extracting the dispersed surface points sets from the simpler shapes, and then compose the points sets through a sequence of new defined Boolean set operation rules. Using these rules, solid object can be defined step by step, and the redundant data can be easily eliminated. In the article, a novel rendering method using the CDSPM medel data is introduced. A re-projection and pixel interpolation method enable us to effectively generate a reasonable image from a set of discrete surface points.

In this paper, we introduce the concepts of multi-node and mono-node to expand the expansive graph grammar given by Shi and Fu to a larger family, the family that contains all rooted acyclic directed graphs. Shi and Fu's approach deals with rooted acyclic digraphs containing only mono-nodes. The ways of numbering and representing rooted acyclic directed graphs are modified and grammars for such graphs are given so that directed graphs containing even multi-nodes can be treated. Parsing algorithms are also given which are described with State-Space representation. Our algorithms are as efficient as those of Ref.(6). We have removed a constraint that the basis graph must be a tree from the expansive graph grammars of Ref.(6) without increasing the space or time complexity of numbering, rewriting and parsing.