We previously developed a signal-processing-based high-resulution image acquisition scheme using two cameras. This approach produces a high-resolution image by integrating two image taken with two cameras. The integration process has two stages : registration and reconstruction. When the two cameras have the same pixel aperture, however, obtaining spatially uniform resolution improvement requires severe limitations both in the arrangement of the cameras and in the geometrical configuration of the scene.
To overcome this disadvantage, this paper introduces two methods : one for reducing noise and the other for improving resulution. The noise reduction method diminishes noise instead of improving resolution by the integration ; the resolution improvement method uses multiple cameras with different pixel apertures. Both methods always provide spatially uniform improvements. We also present an integration algorithm that iteratively performs the registration stage and the reconstruction stage alternately. Furthermore, we introduce a brightness retouch operation that is indispensable in using the alternately iterative algorithm.

View full abstract

Synthesizing left and right images is one of the popular methods used to get a wide-angle image, and this synthesis has conventionally been done by putting two images side by side. The conventional technique thus restricted the arrangement of the cameras and targets. This paper proposes a new method for acquiring a super-wide-angle image by synthesizing images. Left and right rectangular images are transformed into trapezoial images, and the transform formula is derived from relation between the position of the cameras and the target. This method enables the cameras to be set up anywhere and it provides an image nearly twice as wide as the one obtained with a camera, an image easy to see in perspective. This paper also shows the result of a computer simulation for this method.

View full abstract

This paper describes the configuration and performance of a super high definition (SHD) motion picture film digitizing system. A digitizing system for motion picture film and still picture film was developed by using a CCD line image digitizer that can capture images at a maximum resolution of 4096× 4096 pixels, 12 bits/pixel. Motion picture sequences are obtained on a frame-by-frame basis and with an aperture size of 57×57mm from a 70 mm motion picture film. An SHD image processing software enviroment, IMHO, was developed to provide all the necessary image processing functions and a comfortable user interface. A WYSIWYG color matching method, based on direct color measurement/approximation between the original image and the reproduced image developed for the digitizing system, noticably improved color reproduction in a color matching experiment.

View full abstract

Entropy coding algorithms, such as Huffman coding, frequently use bit operations with variable word lengths and although the pipeline processors of existing DSPs can perform numerical processing efficiently, the bit operations in entropy coding are a heavy burden because of a mismatch between these operations and the pipeline processors. We propose a new approach to implement a full programmable entropy CODEC with a combination of a high-level design LSI-CAD system and field-programmable gate arrays. This method simultaneously provides high performance for entropy coding and full programmability. The effectiveness of the proposed approach is demonstrated by using it in designing the Huffman encoder of a JPEG image coding algorithm.

View full abstract

This paper deals with two important problems for 3-D image communication : (i) how to compress the huge amount of data required for the stereoscopic visual effect, and (ii) how to synthesize intermediate-viewing-angle images of a Multi-View 3-D Image. To synthesize a 3-D image having a larger number of views from a 3-D image having fewer views, we first propose a new view interpolation technique in which the resolution of the viewpoint-axis is enhanced for the purpose of increasing the number of viewpoints. Then we present a new data compression method scalable to any resolution ; that is, scalable to any number of viewpoints. For this purpose, the fractal coding scheme is examined. This new method has resulted in the encoding of a 17-view color 3-D CIF image yielding a bit rate of 0.174 bits/pel with a PSNR of 37.9 dB and succeeded in synthesizing 51 views from the compressed data of 17 views, apparently yielding a bit rate of 0.058 bits/pel.

View full abstract

As the first step in the development of a signal compression method for 3-dimensional television systems, the method of searching for disparity between the right and left images has been studied. To increase the accuracy of matching between the distinctive points of the right and left images, two characteristics of the visual system of the human beings have been introduced ; one is the sensitiveness to the contour of the objects and the other is the nonlinearity of the sensitivity to the intensity of the light.
Experiment results show that an optimized matching method called “Rubber-matching” is about nime times more accurate than the conventional block-matching, and about three times more accurate than a high-accuracy matching method using phase correlation.

View full abstract

A method for extracting three-dimensional information from holographic interference fringe patterns by using an interferometer is studied and its effectiveness is confirmed experimentally. The holographic fringe patterns of the object surface, which is assumed to be a set of point sources, are compared with standard patterns calculated assuming that a corresponding point object exists in the measured space. We show theoretically that three-dimensional information can be extracted by three methods : for the sum of products of each pixel, computational correlation using Fourier transformation, and optical correlation. We also simulate extractions of a set of point sources by correlation and optical correlation, and we experimentally confirm the effectiveness of this method applying to actual point-source object and line objects and using interference fringe pattern captured by a CCD camera.

View full abstract

Since the information content of a hologram is so huge, data compression is important for a practical electroholographic display system. We have therefore investigated the block subsampling method and the interpolation of subsampled data. This procedure, however, also causes a blur on the displayed 3D image. Therefore, there is a trade-off relation between the information reduction rate and the image blur. It is not practical to apply 2D image compression to a hologram because the character of the holographic data is quite different from that of a 2D image. We have demonstrated that 2D compression techniques can be used effectively to compress the fringe pattern by first segmenting it and then performing a transform operation upon it. We have discussed both theoretical and experimental results and have compared the image quality of those two compression methods.

View full abstract

Electronic holography is facing two major problems-the huge amount of computation and the poor quality of reconstructed images. We propose a method that reduces the computation time needed to generate holographic fringe patterns by using techniques such as parallel computation with and within many processing elements, vectorized computation in parallel with scalar computation, and computation simultaneous with data transfer to the output device. The result obtained with the proposed method applied to a parallel computer, CYBERFLOW, is 85.4 times as fast as that obtained with a conventional computer and yields a maximum of four frame updates per second. This demonstrates the feasibility of a real-time computer generated holographic display system. We also propose the second problem be solved by synthesis of an electronic hologram and an optical hologram. Experimental results in this paper show that using an edge-illuminated optical hologram for this synthesis provides clear images of both an electronic hologram and an optical hologram to realized a dynamic and precise imaging display.

View full abstract

This paper describes a newly developed volume scanning display that can present a vivid and true 3-D image viewable from any angle. Furthermore, this image can be interactively manipulated by using a 3-D mouse. These characteristics can be implemented with a vertically shuttling LED panel, parallel data read-out circuit, and high-speed DMA data transfer on an Sbus interface. The display can be applied to direct shape modeling, scientific visualization, or medical treatment planning.

View full abstract

Attention is being paid to computer generated holograms (CGH) that can be formed by liquid crystal devices (LCD) because of their possible applications in electro-holography. In the research reported here the kinoform of an on-axis type CGH was constructed by using a commercially available TN type LCD as phase modulator, and the basic characteristics of its image were examined. Moreover an electro-holography system for color 3D animated image display was developed by using three LCD panels, RGB three lasers and graphics frame memory. A clear color image was reconstructed by taking into account the wavelength dependence of the size and position of the color image, then the characteristics of color gamut and animation of the reconstructed color image were evaluated in order to determine the feasibility of this system for electro-holography.

View full abstract

A newly developed, 50-in. diagonal, autostereoscopic full-color 3-D TV display system with a wide viewing angle and requiring no special glasses is described. Most 3-D TV display systems would work with two TV cameras, but we used eight to create actual look-around 3-D images as the viewer moves his/her head. Images shot with eight TV cameras are multiplexed, pixel by pixel, in the signal processing circuit to form a vertically striped image and are rear-projected by two HDTV liquid-crystal video projectors behind a lenticular screen. The resulting image is of high resolution, exceptionally bright, and gives a vivid three-dimensional feeling-and the 3-D effect is obtained from a wide range of viewing angles.

View full abstract

We are developing a Virtual Reality teleconferencing system that offers realistic sensations, and are exploring ways to synthesize real objects and stereoscopic images. We found that viewers felt it difficult to fuse stereoscopic images, even in fusion limit, when they were displayed with real objects. Because there is little information on the fusional limits in situations where real objects and stereoscopic images are synthesized, we examined this difficulty experimentally. The stereoscopic image was displayed on a 70-inch screen, and subjects viewed it by using liquid-crystal-shuttered glasses. The real object was placed to occlude the stereoscopic image. We changed the position and size ratio of these images and asked subjects to rate the difficulty of fusion. The results showed that fusion was extremely difficult when the stereoscopic image was displayed in front of the real object. We also experimentally evaluated situations in which the stereoscopic image moved, and obtained results similar to those obtained when it was stable.

View full abstract

In the perception of binocular stereoscopic images, the Pupper Theater Effect has been pointed out as a problem of size distortion. In this paper, we discuss the effect of pick-up conditions on this problem, especially the effect of a background.
(1) Subjective evaluation tests indicate that a beckground (hallway image) largely affects the perceived size of an object (a manneqin standing in the hallway). They also show that the closer an object is perceived, the smaller the object is perceived.
(2) The effect caused by variation both of focal length of lenses and of interaxial distance between the lenses is examined. The Puppet Theater Effect is more enhanced when the sensation of background depth caused by binocular disparity is dominant rather than when that caused the perspective of a lens used in shooting is dominant.

View full abstract

The accommodative responses of subjects viewing a binocular stereoscopic TV program were measured with TDO III (Three-Dimensional Optometer III). The program was produced with two Hi-Vision cameras and VTR's and converted into NTSC format. A field sequential stereoscopic display with a liquid crystal shutter was used to present the program to four emmetropic subjects. It was found that scenes in 3-D presentation with sudden changes in perceived distance of objects and ones containing objects appearing nearer that the screen cause larger accomodative responses than scens in normal 2-D TV presentation.

View full abstract

The stereoscopic 3D display that uses binocular disparity techniques is widely used as an effective three dimensional display, but there are various problems with stereoscopic 3D images. One of the major problems in stereoscopic 3D display is the mismatch of distance between the accommodation and convergence of human eyes. Experiments were conducted to understand how the stereoscopic 3D images influence human visual functions. The measurement of eye movements showed that cooperative movements of binocular eyes were frequently broken while viewing 3D images in which disparities were large. The measurement of accommodation response time showed that the far-to-near response time was longer after looking at 3D images than it was before viewing. The increases in far-to-near response time increased with the degree of binocular disparity. The results of these measurements showed that within the some limited range of disparity the influence of stereoscopic 3D images may not be significant for human eyes.

View full abstract

Two experiments to examine the effects of the stereoscopic images on the emotional states and on relaxation confirmed the following :
1) Stereoscopic images have both psychological and physiological effects.
2) Psychological effects are more notable than are those of 2D images.
3) Psychological changes depend on images contents (activation and relaxation).
4) Physiological changes (volume pulse wave and skin surface temperature) also depend on images contents.
These results suggest that stereoscopic images cause emotional changes (activation and relaxation) and that there is a possibility of the selfcontrol of autonomic nerves by stereoscopic images.

View full abstract

Because vitrual reality techniques can be used to provide an intuitive user interface using human spatial perception, they are promising methods for implementing a sophisticated user interface. Limitations in computational power, however, make it difficult to present a perfect virtual environment, and a simple task in a real environment therefore often becomes a skilled operation in a virtual environment. This paper describes a method enabling an operator in a virtual environment to place an object on a surface without any force-feedback tool. When a grabbed object comes to close to another object, it is attached. Experimental results show that the method is efficient in decreasing the time required for object placement when precise placement necessary.

View full abstract

In this paper, we describe an integrated driving simulator that enables us to construct a virtual town interactively, and to drive through the town in real time. This simulation system is composed of three functions : road generation, virtual town construction, and driving simulation. We first specify road elements on a graphic terminal, either on the basis of road maps or freely. Then, we construct a virtual town with various elements, such as a road with lanes or signals, buildings, and traffic flow. These elements are automatically generated by using road elements. Finally, we can drive in the vitrual town.

View full abstract