No abstract

No abstract

We present a new video browsing method for multiple videos that are taken at large-scale space for live 3 0 events such as soccer games. By our method, multiple viewers over computer network can browse a live 3 0 eventfrom any virwpoint and each viewer can move hisher viewpointfiee/y Our algorithm consists of five steps. Our system first captures videos from multiple cameras, then rxharts texture segments from the videos, selects appropriate segments according to a viewpoint which is given by user cfvnamically, transmits them to users, and lqouts the segments in virhial space so that each viewer can see the segments in a virtual environment as ifthe viewers were in the event. Dur 3 0 video displq system requires IOMbps at most to browse a soccer game. We conducted experiments at two real soccer stadiums and succeeded in realizing live realistic visualization with fie. viewpoint at about 26fis.

We propose methods for segmenting a motion sequence into motion primitives, taking into account temporal constraints (continuity along the time axis

We propose cone-restricted kernel subspace methods for pattern classification. A cone is mathematically defined in a manner similar to a linear subspace with a nonnegativity constraint. Since the angles between vectors (i.e., inner products) are fundamental to the cone, kernel tricks can be directly applied. The proposed methods approximate the distribution of sample patterns by using the cone in kernel feature space via kernel tricks, and the classification is more accurate than that of the kernel subspace method. Due to the nonlinearity of kernel functions, even a single cone in the kernel feature space can can cope with multi-modal distributions in the original input space. In the experimental results on person detection and motion detection, the proposed methods exhibit the favorable performances.