抄録
H-064
Studies of the Kalman filer and Particle filter method for tracking the moving botanical tree leaves in video sequence.
◎Khanh Do N.・大谷　淳（早大）
Modeling the behavior of a moving non-rigid object such as a moving botanical tree under windy condition is always a tough task to computer graphic because the behavior and structure of the botanical tree is complicated. This paper studies two methods: Kalman filter based method using geometric features and Particle filter based method using depth image data, to track the behavior of a leaf under windy condition. The experiment result shows that Kalman filter based method is fast but not robust;particle filter based method is robust at an acceptable computation cost.