Three-dimensional (3-D) display technology is expected as a future visual information interface, which can provide intuitive understanding for the structure of 3-D information. In addition, Image formation in the air without any apparatus enhances reality for a displayed object and enables a novel visual interface technology that can operate a 3-D image with a pointing device or hands. Since volumetric display technique reported in this report displays an image by forming light points three-dimensionally distributed, the image has natural stereoscopic effect. Moreover, aerial imaging based on retro-reflection is free from distortion. In this report, volumetric display system using a dihedral corner reflector array and a rotating slanted mirror scanner and a technique for suppressing image blur in an aerial image formed by the retro-reflective imaging element are described as examples of researches on areal volumetric image formation.
In the conventional autostereoscopic display technique with a lenticular sheet was developed, as the depth distance of a three-dimensional (3D) image from the display increases, blurring of the 3D image increases. Furthermore, image position changes depending on observing position, and therefore a lenticular sheet is unsuitable to the interactive display, which is operated by visual interface like as touching the aerial image directly. In this report, we propose an aerial 3D display combined a lenticular sheet and a dihedral corner reflector array. This can form a 3D image at a fixed position in the air. Inconsistency between focus and binocular convergence can be alleviated. By combining a motion sensor for human hands, the interaction between the aerial image and the user is allowed, and an intuitive interface can be achieved.
Image inpainting is a method to remove an object in an image and complement the removed area with background image. We present a novel image inpainting method with depth clustering of RGBD image. Image inpainting is based on two technical procedures, first one is estimating a removal area, second one is an image complementation in the area. We use x-means clustering of depth value as estimating removal area. X-means can be applied to a scene which is hard to estimate the number of clusters. We also complement the removed area by using textures which has almost same depth value as background scene.
See-through head-mounted displays (HMD) provide an effective capability for Mixed Reality or Virtual Reality. By using a see-through HMD, an observer can see both real world and virtual world at the same time. However, when two-dimensional virtual information is displayed by a HMD, an observer is hard to understand that it is related with a real-world object. What an observer sees needs to be augmented by 3D virtual information image in accordance with the real object. In this paper, two kinds of head-mounted 3D displays are described. One is a stereoscopic HMD, the other is a super multi-view HMD.
We propose the omnidirectional 3D display system using multiple light field 3D display units which consist of a liquid crystal display and a lenticular lens. The light field 3D displays are arranged in polygonal shape. A 3D image is displayed at center of the polygonal system by using half mirrors. Therefore, observers can see the 3D image around the system. We constructed the prototype system to verify the effectiveness of the proposed system. The prototype system whose radius is 573mm consists of the 4 light field 3D display units which are arranged at 15° intervals. We confirmed that viewing angle is 60° and a 3D image can be displayed with the accuracy of 0.30mm in X-axis, 0.37mm in Y-axis, 1.78mm in Z-axis.