Considering the visual systems of animals including their retinas. it seems essential to know about their ecology, Especially, the relationships between predators and preys are most basic. In this paper, I picked up some physiological and anatomical evidences of animal visual systems, which are not well-known to engineers, and explained them from ecological point. of view. Besides, I refered to some ideas for applications in engineering.

View full abstract

We examined frequency properties and absolute latencies of RGB-type holizontal cell responses in the goldfish in vivo retina. The frequency properties were low-pass filter type with cutoff frequency of 8, 10 and 4Hz, and the absolute latencies were 17, 33 and 52msec for blue, yellow and red lights, respectively. Along with those properties of L-type and RG-type holizontal cells we previously reported, these results support the proposed model of the neural networks between cone photoreceptors and holizontal cells.

View full abstract

This paper describes the spatial contrast sensitivitiles (MTF) of the human eye for stereoscopic displays as a function of stereoscopic depth. Observations shown that there was no significant difference in the MTF as the target depth varied from the display screen position of 0.5 Diop (2m) to 0.07/1.0 Diop. These perceptual data mean that the observer's eyes were accommodated to the screen equally well for all the required convergence distances. On a natural vision, the images of other surrounding objects at different depths from the target are blurred. Accordingly, stereoscopic displays should correspondingly blur such surrounding images for improving the viewing comfort.

View full abstract

Movements of both eyes were measured while the subject was watching Hi-Vision 3D and/or 2D images and fixation points were calculated in a three-dimentional space. Fixation points smoothly moved toward the depth direction following the target's depth movements while the subject watched 3D images. However, for 2D images, the subject's fixation points seemed to move back and forth without following the target's depth movement. Evaluation of 3D vs 2D displays will be possible through the analysis of the fixation points in a three-dimensional space.

View full abstract

Although the CRT displays are generally used to handle the CAD data, it is very difficoult to recognize the images, because the CRTs Can't display stereoscopic images. In the system Proposed here, the images are displayed in front of the CRT, and movement of the hands of the CAD operater is measured in the 3-D display space through two CCD cam-eras. Using this system, the operater can prepare the CAD data as if he is making real objects by his own hands.

View full abstract

3D graphic workstations have a powerful geometric engine to process surface rendering. But using their power to visualize 3D volume data, we have to convert them into surface model. As one of the methods for this conversion, the method of dividing cubic cell(voxel cell) into 5 tetrahedros is useful. In this report, we show the method for constructing 3D cross-sectional image from a set of cross-sectional polygons. The threshold edge of the cross section fits with isosurface. And we have applied this method to molecular graphics modeling.

View full abstract

This report describes a stereoscopic displaying method using holographic stereograms for image data observed by the GMS, namely, the so-called meteorological satellite Himawari. Firstly, an image of the Northern Hemisphere is synthesized from GMS image data. Then, 180-frame images changed a viewpoint little by little are generated fro!n the Northern Hemisphere image data. In the synthesis of a holographic stereogram, these sequential image data are exposed to a holographic film using a liquid crystal television panel as a spatial light modulator. A reconstructed GMS image can be observed behind the film with laser light illumination.

View full abstract

In this paper, an image distribution correction algorithm and its implementation are de-scribed with which we can get a proper contrast image from several contrast-degraded image taken under a back lighting, for example. A principle of conversion described is to measure gray level histograms in realtime then adjust characteristics of analog video amplifiers automatically. Experimental implementations show a good result and prove a correctness of algorithm.

View full abstract

Recently, motion picture coding methods for an ATM (Asynchronous Transfer Mode) networks of a broad band ISDN are studied actively. This paper describes the discussion of the installing the leaky prediction to the motion picture coding method NUCLEI proposed before by the authors. NUCLEI is based on three strategies. The first. NUCLEI uses the interframe coding with the full band image signals. The second, NUCLEI does the interframe coding after the occurring cell loss. The third, the image signals are divided to MSP (Most Significant Part) and LSP (Least Significant Part) by the measure of picture motion, not the frequency. On last proposal, the frame memory refresh method was by use of intraframe coding. But on the more low bit rate coding, the frame refresh method by the leaky prediction is needed. In this paper, the authors showed that the leaky prediction for NUCLEI as the frame memoryes refresh was effective with the discussion and the simulation.

View full abstract

Extensive research and development activities of video coding are currently conducted. Among these activities, fixing total amount of bits per video frame is indispensable for a storage media application. Regarding total bits fixing technologies, coding schemes based on MAX quantization with constant length codes such as Adaptive DCT scheme have been reported up to now, which are far behind of rate-distortion bound. In this paper, bit-rate optimization scheme by trellis coding is proposed. In this scheme. coding performance reaches almost rate-distortion bound using variable length codes, while fixing total generated bits per frame by trellis coding. Effectiveness of this scheme is shown by theoretical analysis and a computer simulation.

View full abstract