We have extended our sports video browsing framework for personal video recorders, such as recordable-DVD recorders, blu-ray disc recorders and/or hard disc recorders, to music segment detection. Our extension to Japanese broadcast music video programs consists of detecting audio segment boundaries such as conversations with guests followed by music/song etc. Our proposed system first identifies the music/song scenes using audio analysis, and then adjusts the start/end position by detecting video shot changes, so as to achieve accurate detection of the music segment thus enabling rapid browsing. Our preliminary results indicate that our audio-only summarization with scene change support works well for music video content. We can therefore integrate the enhancement into our product at a low computational cost. IEEE Transactions on Consumer ElectronicsThis work may not be copied or reproduced in whole or in part for any commercial purpose. Permission to copy in whole or in part without payment of fee is granted for nonprofit educational and research purposes provided that all such whole or partial copies include the following: a notice that such copying is by permission of Mitsubishi Electric Research Laboratories, Inc.; an acknowledgment of the authors and individual contributions to the work; and all applicable portions of the copyright notice. Copying, reproduction, or republishing for any other purpose shall require a license with payment of fee to Mitsubishi Electric Research Laboratories, Inc. All rights reserved.

We have developed an aerial display that people can pass through a 56-inches floating image as if they dive into it. We used a curved retro-reflector so that a reflection image which overlapped with a floating image disappeared. Guide images help to focus viewer's eyes on the blurred floating image.

In this paper, we propose a new method that makes multi-layer images from a few viewpoint images to display a 3D image by the autostereoscopic display that has multiple display screens in the depth direction. We iterate simple "Shift and Subtraction" processes to make each layer image alternately. The image made in accordance with depth map like a volume slicing by gradations is used as the initial solution of iteration process. Through the experiments using the prototype stacked two LCDs, we confirmed that it was enough to make multi-layer images from three viewpoint images to display a 3D image. Limiting the number of viewpoint images, the viewing area that allows stereoscopic view becomes narrow. To broaden the viewing area, we track the head motion of the viewer and update screen images in real time so that the viewer can maintain correct stereoscopic view within +/-20 degrees area. In addition, we render pseudo multiple viewpoint images using depth map, then we can generate motion parallax at the same time.