As information support to deaf and hard of hearing people who are viewing sports programs, we have developed a sign language support system. The system automatically generates Japanese Sign Language (JSL) computer graphics (CG) animation and subtitles from prepared templates of JSL phrases corresponding to fixed format game data. To verify the system's performance, we carried out demonstration experiments on the generation and displaying of contents using real-time match data from actual games. From the experiment results we concluded that the automatically generated JSL CG is practical enough for understanding the information. We also found that among several display methods, the one providing game video and JSL CG on a single tablet screen was most preferred in this small-scale experiment.

To expand the services that are based on Japanese Sign Language (JSL) for deaf and hard of hearing people, we developed a support system for viewing sports program. The system provides sign language computer graphics (CG) animations and other auxiliary information such as text, image, and notifications automatically generated from game metadata. Results obtained from gathered opinions showed that the system is effective for understanding the situation when a game is interrupted.

Context-awareness provides users with more versatile user interface environments when using digital devices. In a TV viewing environment, context-awareness techniques can be used for recommending useful information related to the viewed program. To incorporate a user's preference in doing so, it is vital to infer whether the viewer is actually interested in the TV program. Therefore, we propose a method for inferring viewer interest in a TV program. In the proposed method, we regard the problem of inferring viewer interest as a sequential labeling problem and solve it by applying latent dynamic conditional random fields to data sequences generated by integrating the operational log information of user interaction and visual information of viewer behaviors.

We have developed a robust method for estimating camera parameters for live TV production based on infrared markers whose feature-point extraction is simple and two-dimensional color histogram matching that considers the effects of specular reflection due to lighting. We first explain the principle of the proposed technique. We then present the results of a basic experiment for evaluating the accuracy of marker identification and the results of an image-compositing experiment using estimated camera parameters. We show that the proposed technique is effective based on those results.

When making a TV program in a studio, care must be taken that the camera does not shoot beyond the boundary of the studio set. In addition, limitations in cost and space for the set must be taken into account. In a virtual studio, on the other hand, we can solve this cost and space problem, but in turn, actors are requested to perform in front of a blue background screen, which is not always an easy task for them. To solve these problems associated with real studios and virtual studios, we have developed a new type of virtual studio called Real Space-based Virtual Studio in which a real space image and virtual space image are combined naturally with no boundary seam. There are two major advantages in using this new virtual-real hybrid system. One is that the actors can concentrate on their role in the real studio sets, and the other is that camera work can be done without worrying about off-screen areas of the set. In the present study, we constructed an omnidirectional image with ultra high-definition features and combined it as a virtual studio image with a real studio image. We have developed an integration system and from the experimentation we have shown that the omnidirectional images and the real studio images can combine smoothly and naturally.

We are studying an image-based virtual studio. This type of virtual studio makes it possible to reconstruct realistic synthesized video by extracting shape, texture, and other necessary information (called video components) from each object in the subject video for use on the set. When using video components made up of shape and texture information to reconstruct natural-looking video under any lighting conditions, lighting effects like highlights and shading that are present at the time of measurement must be removed from texture information. For this purpose, the authors have proposed a technique that uses polarized light on the lighting side and a polarized filter on the camera side, and have confirmed its effectiveness by experiments.