Ken'ichi Kakizaki

Generating the animation of a 3D agent from explanation text

Thk paper proposes a presentation agent designed for a virtual environment The agent, animated in the virtual environment, performs a presentation based on a speech teti that e~~lns objects in the environment. h order to gene~te the agent’s motion automatim~, we categotie the presentation motions of the agent into three classes, pointing, moving, and gestiring. Based on this =tegotition, our ystem etiracts words from the eq~nation teti. In order to determine what the target object k and to efiract detatied information about the object the ~stem accesses a scene gmph that contains an the information about the vitial environments. The ~stem wn find the object based on its charactetilcs that are eqlained h the teti, and can then efict information about the object such as its S* position, and weight. This information is used for determining detaik to generate the agent’s motion. 1.1 Ke.words

Suitable representation of values on the logistic map with finite precision

We are investigating the logistic map over integers and its pseudorandom number generator. Focusing on differences between the logistic maps over floating-point numbers and over integers, we herein show that the integer format is a suitable representation for implementing the logistic map on computers with limited memories.

A terminal authentication with transiting nodes by SymPC on an ad-hoc network

In this paper, we propose an one-sided authentication method by the SymPC which is a homomorphic symmetric encryption in an ad-hoc network such as a sensor network. In our protocol, a manager can authorize a node and know transiting nodes between them by using secret keys of their nodes.

Image segmentation methods for sanitizable signatures

In the presented paper, we give two feature-based segmentation methods of a digital image such as a photo or a scanned document for sanitizable signatures. Since the image must be partitioned into multiple segments before one is signed, we focus on features such as faces in photos and characters in scanned documents. Our methods divide the photo image into segments with the same pixel value and the scanned document into segments with a character. We also proposed a sanitized signature scheme that can detect forged segments if the signature is invalid.

A study on distance between two sequences by the logistic map over integers with slightly different parameters

One of our investigation purposes is to give a design guide of a pseudorandom number generator using the logistic map with finite precision. In the present paper, we show that two sequences for slightly different parameters are completely different from each other, even if the logistic map has the same computational precision and the same initial values. First, we decide the arithmetic precision and the initial value. Second, we obtain two sequences by the logistic map for slightly different parameters. Last, we estimate a correlation of them by measuring the distance between two elements.

Incremental Scene Graph Distribution Method for Distributed Virtual Environments

This paper describes an incremental scene graph distribution method for client-server based distributed virtual environments. It is very important for distributed virtual environments that all users share the same environment on a service. In order for users to share an environment, a scene graph that describes the virtual environment must be distributed to all clients by the server. The method we propose divides a scene graph into segments and distributes these segments incrementally. The distribution is performed according to the importance of the objects visibility, and the measurement of importance is evaluated by each client. This measurement is performed in a traversing process for scene rendering, so it does not expend any extra processing cost. If an important but still as yet undistributed scene graph segment is found by a client, the client requests its distribution of the server, at which time the server distributes the segment upon demand. The method realizes an efficient scene graph distribution in a narrow bandwidth network.