@@ By superimposing some additional information(e.g. ball trajectory in baseball game and visualization of offside line in soccer game), a sports broadcast has become much attractive for audiences. For creating such additional information for many sports, it抯 necessary to track the objects (e.g. players or balls) steadily. Particle .lter [1] based on prediction with many distributed particles is well known as a robust object tracking algorithm and can be a promising tool for it. However, since conventional methods use one state transition model, the tracking accuracy is decreased for the objects with irregular motion which is often seen in many sports.
展开▼
