Three-dimensional integral television using high-resolution video system with 2000 scanning lines

Abstract

Use of three-dimensional (3-D) images in broadcasting, communications, and many other areas has been anticipated for some time, but practical applications have shown little progress. One reason is that most 3-D imaging systems fail to simultaneously provide: (1) binocular disparity that can be experienced without special glasses, (2) a convergence point that matches the eye's accommodation point, and (3) motion parallaxes that enable an observer to see different images corresponding to different positions horizontally and vertically (full parallaxes). These capabilities would enable observers to see a 3-D image as though it were a real object. By expanding research on spatial imaging, we aim to develop a 3-D imaging technology that performs all these functions. Depending on the characteristics of the reconstructed images, 3-D systems can be roughly classified into four types: binocular, multi-view, volumetric imaging, and spatial imaging. The spatial imaging type is able to satisfy the three requirements mentioned above. Holography is a well-known example of this type. The integral method is also a spatial imaging type. Because it can produce 3-D images using natural light (incoherent light), it is considered to be one of the ideal 3-D systems. The integral method was first proposed by G. Lippmann [1] in 1908 based on a photographic technique. Recently, several attempts have been made to obtain higher quality [2, 3, 4, 5, 6, 7, 8] and moving 3-D images [9, 10]. Figure 1.1 shows the principle of the basic integral method using a single lens array in the capture and display stages. To produce an integral image, a lens array composed of many convex elemental lenses is positioned immediately in front of the capture plate. The integral image is composed of numerous small elemental images that are captured and recorded on the plate. The number of images corresponds to the number of elemental lenses. The integral image is supplied to a transparent display plate. The display plate is placed where the capture plate had been, and is irradiated from behind by an incoherent light source. The light beams passing through the display plate and the lens array re-trace the original routes and then converge at the point where the object had been, forming an autostereoscopic image. The total number of pixels, Nt , on the capture plate is the product of the number of elemental lenses, Nm, and the number of pixels, Ne, in each elemental image horizontally and vertically: Nt = NmNe. (1.1) The number of elemental lenses determines the upper limit of resolution. The number of pixels in each elemental image affects the resolution of the 3-D image away from the lens array. The capture plate requires Ne times the number of pixels required by conventional television. Thus, extremely high resolution is required for both the capture and display plate. We experimentally developed a 3-D integral television using an extremely high resolution video system with 2,000 scanning lines that provides full color and full parallax 3-D images in real time. © 2009 Springer-Verlag New York.