A multi-view 3D video processor was designed and implemented with several FPGAs for real-time applications and a projection-type 3D display was introduced for low-cost commercialization. One high resolution projection panel and only one projection lens is capable of displaying multiview autostereoscopic images. It can cope with various arrangements of 3D camera systems (or pixel arrays) and resolutions of 3D displays. This system shows high 3-D image quality in terms of resolution, brightness, and contrast so it is well suited for the commercialization in the field of game and advertisement market.

The authors developed a simple stereoscopic display using the characteristic of inclined polarization of existing LCD panels. The production time of this display is very short and the cost is low price. The screen size of this display becomes about half of the LCD panel. Thus the stereoscopic images are displayed on the slightly small screen. Therefore several distortions often appear as puppet theater effect and cardboard effect. The researchers have not been able to provide the satisfactory solutions to these distortions. The authors propose a geometrical model to describe the relation between the real space of the recording time and the virtual space of the replay time. This model is mainly related to the reduction or the magnification about the screen size and the distance of a pair of cameras. The authors could improve the distortions by considering these conditions. Moreover the authors succeeded in improving 3D images more vividly. The essential point is the construction of multiple layered virtual images. The authors call this complex image "sur-virtual image." In this way the volume of the expressed object and the scale of the virtual space can be easily expanded. The viewer can enjoy more exciting 3D images.

Previous research from our laboratory indicated that sense of presence is enhanced for stereoscopic images with scenes typical of video telephone (VT) systems, compared to non-stereoscopic 2-D images. However, this enhancement was not found for all sequences. In the present study we report results obtained with a new set of stereoscopic sequences containing depth scenes that were created through manipulation of camera focal length, background scene, and camera convergence point. Viewers were asked to rate both stereoscopic and non-stereoscopic versions of the sequences on naturalness and sense of presence. The methodology of double-stimulus, continuous quality scale (ITU-R Recommendation 500) was used in the subjective assessment. Images in the video sequences were common image format (CIF, 352 x 240 pixels) with a display size of 15.5 cm x 11.6 cm. The results confirmed our previous findings that sense of presence is enhanced for certain stereoscopic video sequences, compared to non-stereoscopic sequences. The results also indicate a high correlation between ratings of naturalness and sense of presence (r² = 0.75), although ratings tended to be lower for naturalness than for presence. Both ratings tended to improve slightly with camera focal length, except for sequences with a natural background. For the range studied, no effect of camera convergence point was found.

This paper describes a new means to control depth positions of 3-D images for integral photography. A GRIN lens array is set in front of a lens array for image capturing. The length of each elemental GRIN lens composing the array is a half of one period of the optical path. The GRIN lens array also avoids pseudoscopic effects with revised depth. The depth position of the 3-D images is controlled by adjusting the distance between the GRIN lens array and the capturing lens array, thus producing 3-D images without depth distortion.

This paper describes an arbitrary viewpoint 3D display system that is a new concep of a 3D display. This system consists of a multivie camera system for 3D data capturing and a real-time color autostereoscopic display system to show an arbitrary viewpoint 3D image. This 3D display system is based on the reconstruction of parallax rays, and it is one of real-time integral imaging systems. To reconstruct a 3D image, the ray-space method is used. By using similarities between the reference plane in the ray-space representation and the screen of the 3D display system, the ray data to display a real 3D object can be obtained with multiview images. Moreover, an arbitrary ray data can be synthesized from captured multiview images. Therefore, an arbitrary viewpoint 3D image can be reconstructed by the 3D display system. This paper describes the principle of an arbitrary viewpoint 3D display system, and also describes the experimental results.

Three dimensional (3-D) displays play an important role in the field of entertainment. Today, research is being conducted to produce 3-D displays to meet the complex needs of high-functionality full motion 3D displays at reasonable cost, but without glasses, complicated viewing arrangements or restricted fields of view. Other applications for 3-D displays include but are not limited to CAD/Design simulation, advanced data representation, displaying complex 3-D information for automotive design, medical imaging, advanced navigation displays, scientific visualization, and advertising. The key element in all these applications is an optical beam scanner that can display 3-D images for large viewing angles. Our proposed Code Multiplexed Optical Scanner (C-MOS) can fulfill all these requirements with its high beamforming power capabilities. Our proposed experiment demonstrates three dimensional (3-D) beam scanning with large angles (e.g., > 160°), large centimeter size aperture, and scanning speed of <300 μsec. The robust construction and simple operation of the C-MOS makes it very useful and attractive for deployment in the field of entertainment, defense and medical imaging. Here we report the application of the C-MOS for three dimensional (3-D) displays.

This paper deals with 3D object classification using computational holographic imaging. A 3D object can be reconstructed at different planes using a single hologram. We apply Principal Component Analysis (PCA) and Fisher Linear Discriminant (FLD) analysis based on Gabor-wavelet feature vectors to classify 3D objects measured by digital interferometry. Experimental and simulation results are presented for regional filtering concentrated at specific positions, and for overall grid filtering. The proposed technique substantially reduces the dimensionality of the 3D classification problem.

A three-dimensional (3D) object transmission system by use of digital holography is presented. We can combine holography to detect 3D object information with a pulse shaper based on ultrashort pulses for ultrafast data transmission. A 3D object is sensed by a fast medium using digital holography. Then, the output of this medium is interfaced with an ultrafast optical system. After the data transmission, the information is reconstructed optically in a 3D form using a display device. System configuration and numerical evaluation are presented.

Digital holograms are capable of encoding three-dimensional (3D) information of a scene or object. Uniform quantization has been used in the past to compress digital holograms with some success. We have developed nonuniform quantization techniques that operate on the complex-valued hologram data for increased performance. Our digital holograms of 3D objects are captured using phase-shift digital interferometry. The quantized holographic pixels are coded using lossless techniques to achieve compression rates in excess of 40.

Fundamental experiment of holographic memory was carried out for university students to understand thebasic concept of the memory as a large format storge method of next generation. We performed holographic data recording experiment using simple two beam optical systems. Multiple recording to enhance storage capacity was also successfully performed.

Free-viewpoint TeleVision (FTV) is a next generation television where users can move their viewpoints freely. In the previous paper, we reported an FTV system based on the Ray-Space representation. In this paper, we focus on acquisition and display system for the FTV. As an acquisition system, we investigated two configurations: (1) multiple cameras with interpolation, and (2) a single high-speed camera with optical scanning system. As a display system, we developed a display with head tracking, where the position of a user is detected by image processing.

To allow multiple viewers to see the correct perspective and to provide a single viewer with motion parallax cues during head movement, more than two views are needed. Since it is prohibitive to acquire, process, and transmit a continuum of views, it would be preferable to acquire only minimal set of views and to generate intermediate images by using the estimated disparities. For high quality of the generated image, the first, we propose how to generate the intermediate images using multi-resolution and irregular-quadtree decomposition. Irregular-quadtree decomposition is aligned at the object boundary which is the disparity discontinuity. By finding the peak over the absolute values of the high pass filtered output that is applied to the row and column average, the horizontal and vertical dividing locations of the block are computed. The second, regions of occlusion are decided by similarity comparisons among the matched block alternatives, then filled with the pixels of left or right image by the principles we proposed. Finally, the images at arbitrary viewpoints of generated and yielding a 31.1 dB PSNR at middle location between both viewpoints.

In the field of 3-D imaging technology, Integral Photography (IP) is one of the promising approaches, and a combination of an HDTV camera and an optical fiber array has been investigated to display 3-D live video sequences. The authors have applied this system to a computer graphics method for synthesizing arbitrary views from IP images: a method of interactively displaying free-viewpoint images without physical lens array. This paper proposes a real-time method of estimating depth data corresponding to each element image on an IP image. Experimental results show that the proposed method is very useful for improving the quality of the free-viewpoint image synthesis.

We propose an optimum nonlinear filter to detect the target's 3D coordinates within the input 3D scene using LADAR data. The 2D encoded LADAR range image is converted into 3D binary profile, and then the 3D optimum nonlinear filtering technique is used to detect the 3D coordinates of targets including the target distance from the sensor. The 3D optimum nonlinear filter is designed to detect distorted targets, i.e., out-of-plane and in-plane rotations and scale, and to be noise robust. The nonlinear filter is derived to minimize the mean of the output energy due to the disjoint background noise and additive noise and output energy due to the input scene, while maintaining a fixed output peak for the members of the true class target training set. The system is tested using real LADAR imagery in the presence of background clutter. The correlation outputs of LADAR images show dominant peaks at the target 3D coordinates.

We present a method to compute high-resolution three-dimensional (3-D) images based on integral imaging. In order to overcome optical limitation on the spatial resolution of reconstructed 3-D image, we suggest performing time-division multiplexing by which a sequence of II's is captured with a slightly different location of the lenslet array. The II image sequence obtained by the detector array is processed digitally with super-resolution reconstruction algorithms to obtain a reconstructed image, appropriate to a viewing direction, which has a spatial resolution beyond the optical limitation.

Integral images contain three-dimensional (3D) information about a scene. We extract this information using a stereo-matching algorithm and we digitally reconstruct the 3D scene in a computer. We then use the reconstructed scene to perform 3D recognition by means of a nonlinear 3D correlation. We demonstrate the recognition and localization of objects in a 3D scene. We also compare discrimination of 2D and 3D correlations.

In this paper, a new web-based multi-view 3D display system through the intermediate views synthesis using an adaptive disparity estimation algorithm is proposed and its performance is analyzed by comparison to those of the conventional disparity estimation algorithms. First, the feature value indicating the local image complexity is extracted from the input stereo sequences by using the edge detection and disparity estimation. Then, the matching window size for disparity estimation is adaptively selected depending on the magnitude of this feature value. This new approach is not only able to reduce the disparity vector mismatch encountered in the conventional dense disparity estimation with small matching window size, but is also able to reduce the blocking effect associated with the disparity estimation using a large matching window size. With some experimental results on a sequence of stereo image captured by the web-cameras, it is found that the adaptive algorithm improves PSNRs on the average in comparison with those of the conventional algorithms and multi-view images can be generated through the process of real-time IVS. Finally it is suggested that this approach can be applied to the web-based real-time 3D video communication system.

In an effort to make automatically detect image features for pattern recognition, we described a 3-dimesional (3-D) Hough transform. We describe two interlocking theoretical extensions to greatly enhance the Hough transform's ability to handle finite lineal features and allow directed search for various features while balancing memory and computational complexity. We computed the 2-D Hough transform of 1-D slices of an image which results in a 2-D to 3-D transform. Features such as line segments will cluster in a particular location so that both line orientation and spatial extent can be determined. This approach allows the Hough transform to be more widely applied in pattern recognition including 3-D features.

A natural 3D image can be generated by projecting massive 2D images. A number of afocal and telecentric imaging systems are aligned in a modified 2D array and are multiplexed to project a number of 2D images into different horizontal directions. Such multiple imaging system has large aberration. The combination of lenses is used to reduce the aberration of a Fresnel lens which is used as a common lens of the multiple imaging system. Electronic correction is also used to reduce the image distortion. These two methods greatly improve the 3D image quality.

Natural 3D images can be generated by displaying high-density directional images. In order to display real 3D scenes, a 3D camera technique which generates high-density directional images of real 3D scenes is required. The most promising method is to generate them by using an interpolation algorithm from multiple images captured by horizontally aligned multiple cameras. In this study we examined four different simple interpolation algorithms. The algorithm that utilizes one representative object distance as a priori was found to be the most effective one. This technique offers the fast interpolation. However, there is a tradeoff between the number of cameras and the depth of 3D objects. The allowable depth for 3D objects is reported.