We assessed the detectability of a small dot target presented in half-occluded region of three dimensional images. The threshold detectability of the target which was superimposed for about 67msec on half-occluded region was higher than that which was on two dimensional region. This result suggests that the alternative changes of dominant eye frequently occur in half-occluded region, and they affect the perception of sharpness and quality of three dimensional images.

View full abstract

The relationship between the subjective sizes of two stereoscopic targets and the vergence of subjects viewing the targets were studied. On the condition that the disparity between the two targets were smaller than Panum's fusion area, the ratio of apparent size of the target to the real size was consistent with the geometrically calculated value. On the other hand, when the disparity was too large for the subjects to fuse the two targets simultaneously, the ratio was not consistent with the calculated value. It was found that this inconsistency under the large disparity condition can be explained by the change of correspondence between the retinal images caused by the displacement of the vergence point.

View full abstract

The relation between the timing of blinks and gazing-objects while performing visual task is analyzed. In the experiments, capital and small letters of alphabet are presented to a computer display in horizontally distant position. Under the counting conditions, the subjects are instructed to count the total number of the target characters, while they are instructed to merely watch the screen under the free-observing conditions. As the results next findings are obtained. In the formar conditions, the blinks triggered by gazing-letters onset are frequently occurred and the blinks with slightly longer latency tend to occur in association with the target rather than with non-target characters onset. In the latter conditions, the blink onset in association with either of capital or small letters are likely to be observed continuously although gazing-points frequently go back and forth between letters.

View full abstract

Real-time input devices of human body location and motion are important in virtual environments. In this paper, an optical 3D position detection system using a pair of optoelectronic semiconductor position sensing devices as real-time input device is mentioned. 3D positions of light emitting markers are derived by the principle of stereo photography, and an experimental system detecting multiple light emitting markers is implemented with simple electronics and a personal computer. Some potential applications suitable for this system in virtual environments are also proposed.

View full abstract

In the field of ophthalmology, it is often required to observe three-dimensional structures around an optic disc for early diagnosis of glaucoma patients. Using a newly developed Confocal Laser Scanning Ophthalmoscope (CLSO), we can obtain a series of images with focus positions at slightly different depths. A method of reconstructing three-dimensional structures from a series of CLSO images is proposed. That is, three-dimensional structures are reconstructed by using in-focus areas extracted from the series of CLSO images. For accurate reconstruction, even when the number of images are small, in-focus areas are determined by interpolation. The Usefulness of the proposed method is demonstrated with pan-focused stereoscopic displays of retina.

View full abstract

We have developed a new video composite system with a motion control camera and a real-time video processor. The system can realize dynamic and seamless camera work with wider range from long distance to close up. Within movable area of the motion control camera, it moves actually. The camera moves out of it, the image shot by the camera is transformed geometrically by real-time video processor, as if the image is shot from virtual camera position.

View full abstract

In this paper, described is an empirical study on a virtual object manipulation in a shared virtual environment. "Virtual object hand-over task" and "Virtual object joining task" are studied respectively. In the hand-over task, task completion time was collected over four different visual feedback conditions. The results clearly shows that the visual feed back which clearly shows the object holding status are important for the better task performance in time and an error rate. In object joining task, effectiveness of "collision avoidance" and "d.o.f.(degree of freedom) constrains" were tested. Task completion time joining two virtual objects was measured. Our results shows that collision avoidance method are more effective against d.o.f. constraints method.

View full abstract

Research on shooting methods of 3-D program production for natural and fatigue-free 3-D images has been continued. Toed-in and parallel camera configurations are available for shooting 3-D programs. The former is usually used because shooting and viewing conditions can be simply set to get a desired 3-D design. It has been shown, however, that this camera configuration brings about the inconsistency between depth information by lens perspective and that by binocular parallax, and that the inconsistency leads to the size distortion called "Puppet Theater Effect".

View full abstract

The Hi-Vision Theatre which is placed inside the NHK-Hirosima Broadcasting Center Building is a multi-purpose theatre with a 200-inch rear projected HDTV screen that can project bright and 3-dimensional pictures. This is a report of the outline and the yearly record of the usage.

View full abstract