Research of trajectory tracking in virtual sculpture based on machine vision

For the low accuracy and limit application feature of force feedback devices and data gloves tracking in the virtual sculpture system, the paper has proposed a target tracking algorithm which uses machine vision in the virtual sculpture system for the first time, making the application range of virtual sculpture expand greatly. The algorithm is based on Kalman filtering, includes a method of dynamic determining search area to narrow the search range for real time and improves the seed filling method, so that the computation and redundancy of target searching can be reduced greatly, which makes the accurate positioning more efficiently. The combination of the above three methods makes tracking speed be greatly enhanced to achieve an accurate real-time trajectory tracking.