In this paper we propose the LED array projector that project invisible information for small embedded device. Each pixel of this projector has own LED light source which can easily modulated by high frequency signal. In this system, projector send the image to the human's eye and invisible information to the embedded device. For example, each pixel can send positional information and different video signal. We develop a proto-type projector which has LED array of 5 pixel width and 4 pixel heights.

抄録全体を表示

In this paper, we propose the system replacing half-mirror in the RPT (Retro-reflective Projection Technology) system by micro full-reflection mirror. This system enables to project wide-field image without covering user's eyes. Conventionally, several Head-Mounted Displays are developed as AR (Augmented Reality) display. However, most of these displays cover an user's eyes, and consequently generate low realistic sensation. We present novel head-mounted projector "X'tal Visor" based on this system, and conduct series of performance evaluation tests.

抄録全体を表示

This paper proposes a practical fully-automatic method of 3D modeling for VR-systems by utilizing moving shots of a stereo camera. VR-modeling based on measured data is generally composed of the following process: 3D measurement with high-precision laser range sensors, texture acquisition with high-definition digital cameras, and alignment between 3D model and texture. However, this is often the case laser range sensors are difficult to handle with and the alignment process is mostly dependent on manual work by trained operators. To overcome the above problems, we give an eye to the stereo camera, which is handy to use and can simultaneously acquire the 3D shape and texture, though its accuracy is not as good as that of laser sensors. And utilizing the feature tracking technique on the image sequence, we can automatically align all the partial models estimated from stereo pairs and observe the whole model immediately after the shooting. Our system enables automatic generation of VR-model by simply moving a stereo camera around a target object and taking a series of stereo pair images. Finally, we show several kinds of VR-model generated by our system and confirm that our system is effective for a specific application, such as a VR-contents presented in the theater.

抄録全体を表示

The Horizontal fixed viewpoints Biconical Paraboloidal:HBP mirror is an anisotropic convex mirror that has inhomogeneous angular resolution about azimuth angle. In this paper, we propose a remote surveillance system with a specially designed HBP mirror system for collecting information of devastated areas. We used a virtual remote surveillance environment for investigating effectiveness of the HBP mirror system compared with conventional omnidirectional imaging systems. From results of object searching experiments, we confirmed that objects can be detected early and with certainty by the HBP mirror system. We also developed a real remote surveillance system for actual surveillance experiments. Through those two types of experiments, we got the results that the HBP mirror system effectively works for remote surveillance.

抄録全体を表示

Many kinds of tracking for Augmented Reality had been proposed. In case of the feature point tracking, the pose is computed by minimizing the error between the observed 2D feature points and the back-projected feature points from the 3D scene model. This minimization problem is usually solved by a non-linear optimization. The main advantage of this approach is its accuracy. However, it is difficult to compute the correct pose unless an appropriate initial value is used. In addition, when some errors are included in the observation, this approach does not guarantee the correct pose even if it converged on the global minimum. So, once an incorrect pose was computed in one frame, tracking may fail in the next frame or the result will get farther from the correct one. In this paper, we propose a new tracking framework for augmented reality. Proposed method tracks features as multiple local hypotheses based on not only one pose but also multiple poses that are computed in the pose estimation in the previous frame. Since multiple poses are maintained as global hypotheses, as far as the correct pose is contained in the hypotheses, the tracking can be continued in even hard situations like a simple iterative scene with high-speed movement.

抄録全体を表示

We propose a system of AR display for strategy based pool supporting system which analyzes the ball arrangement. First, we estimate a position of each ball from images captured by moving a camera around the table. Then, we classify the balls by ball number. Finally some desired ball paths are displayed on a LCD screen of the PC to support the player's strategy. In our method, captured images are divided into a green table region, balls and others. To estimate the position of each ball on the table, we calculate a homography between a table in the image and an actual table. Then, we estimate a projection matrix based on the homography. Finally, we estimate 2D ball positions on the table by the projection matrix, and classify the balls by voting method. After analyzing the ball arrangement, we show the motion of the balls in the captured image by using CG. In the experiment results, we evaluate the accuracy of the position estimation of the balls.

抄録全体を表示

This paper proposes an automating method to determine a viewpoint for generating a 3D video of soccer games. Our proposed method aims to realize an empathic 3D video system that counts users' personal preference. Our system first classifies the soccer game into some soccer events by referring to 3D position of soccer players and a ball. For each event in soccer games, typical viewpoints, which are preferred by many numbers of users, have been estimated by a clustering method with interview survey. In a phase of presenting 3D video, the user inputs his/her individual taste of viewpoint by trial-using 3D video system in a few soccer events. Our system estimates a viewpoint to generate the user's favorite 3D video by choosing a typical viewpoint which is the nearest to user's input. With interpolating the selected typical viewpoints in soccer events by B-spline curve, this system generates a sequence of 3D video. We confirm the effectiveness of our proposed method in subjective evaluation experiments.

抄録全体を表示

A mixed-reality face (MR face) is a mosaic face with real and virtual facial parts, presented by overlaying a virtual facial part on a real face using mixed-reality techniques. An MR face is an effective means to improve communication in mixed-reality space by restoring the eye expressions lost when wearing HMDs. Photometric registration between the real and virtual parts is important because our eyes are very sensitive, even to small changes in human faces. However, efforts to achieve perfect 'physical' photometric registration on an MR face are not feasible in an ordinary MR space. Therefore, it is essential to clarify the sensitivity of our eyes to the photometric inconsistencies on an MR face, and to concentrate on resolving them. In this paper, we first present the results of a systematic experiment that evaluated our sensitivity to the photometric inconsistencies on an MR face. Then, a technique to resolve the inconsistency and an experimental system to demonstrate the effectiveness of an MR face are described.

抄録全体を表示

Sound field reproduction based on wave field synthesis is a technique that synthesizes wave fronts in a listening area using multiple microphones and loudspeakers placed at the boundary of the area. This study evaluates the effects of the directivity of microphones and loudspeakers on the accuracy of synthesized wave fronts. Microphones in an original field and loudspeakers in a reproduction field were placed at the boundary of both areas which have two cases (a circle with a radius of 2 m and a square 4 m on a side). Three directional patterns (omnidirectional, unidirectional, and shotgun) were designed as directivity conditions of microphones and loudspeakers. The results of computer simulation show that there is almost no effect due to directivity of loudspeakers and that unidirectional or shotgun microphones can reproduce accurate wave fronts.

抄録全体を表示

We propose a modeling system that enables users to create tree models with 3D gesture input. It generates tree models by using growth simulation based on the trunk or silhouette shapes of trees given by user gestures. The system carefully addresses the fragile balance and tradeoff between the freedom of user interaction and the autonomy of tree growth. Users intuitively and easily create tree models that have the exact features of branching structures or the silhouette shape of trees according to user intentions and imagination.

抄録全体を表示

This paper presents a method of interactively generating whole body animation using reduced dimensionality from multiple captured data sequences of whole body motions. The proposed method is achieved by two techniques; automatic motion segmentation and estimation of the posture using multiple regression analysis. Even when the skeletal structure of the user who inputs the motion is different from that of the shape model in the computer, the motion that the user imagines is generated. The proposed method generates animation naturally, just as users would expect.

抄録全体を表示