This paper describes the influence of delay time on manipulating object in a virtual space. Two different experiments have been done to investigate such influence. The first is that the degree of the feeling on motion parallax's delay is experimented by questionary. Its result shows the limit time when we don't feel its delay is below 120 msec. The second is the time measurment of geometical matching task between two computer graphics' object. Its result shows the limit below which delay time has no effect on users' task is less than 120 msec.

View full abstract

A stereoscopic display and dissection method of brain tissue is developed for the easy understanding of its anatomical domains with their relative positions and dissection surfaces. A set of transverse section frame lines obtained from photos of rat brain slices is composed into a three dimensional image. It is viewed as a three dimensional image by adopting stereoscopic view method for giving a depth of synthesized image of rat brain, making out two different images taking into account a human binocular parallax. A dissection part is given in a plane and by polygon using a mouse device and displayed stereoscopically. Furthermore, using a operational knife graphically synthesized with three dimensional position given by a mouse device, a dissectionpart is obtained with its stereoscopic view making clear a trace of its blade of a simulated knife.

View full abstract

In this paper we describe a subband coding of images which uses Extended QMF. Extended QMF partitions a signal into an arbitary number of subband signals. In other words, the number of bands, into which the signal is partitioned, is not constrained to power of two. Moreover we can advantegeously convert the sampling frequency of the singal. Application of Extended QMF to picture coding makes the conversion of image size easier. By partitioning an image into several subband images and transmitting each subband image independently, recievers will be able to select resolution and recieve images at the size of their preference.

View full abstract

Pedestrisan information, which implies the number, position, speed, and direction of them, is needed for management of public facilities. From this information, we obtain traffic density and coefficient of utilization of facilities in detail. In case of emergency, it can offer the information for evacuation. But recently counting of pedestrian has been acheived by hand or special gate. The purpose of this study is to propose two methods : 1) Monitoring of pedestrian by image processing, and 2) Removing shadow of pedestrian in image. The method 1) is performed by using serial difference of image, we can get the number and position of pedestrian, and from sequencial image data we reckon their speed and direction. The shadow of pedestrian makes error in counting number of them, therefor method 2) is needed to obtain accurate number. In conclusion, we can remove the shadow, and if the number of pedestrian is under 20 per 10 meters around, we can trace each pedestrian.

View full abstract