Moving object tracking research based on active vision
Feng Pan, Wang Xuanyin, Quanqiang Wang · 2004
In an active vision condition both camera and object may move simultaneously and common motion detection methods could not deal with such situation. To overcome the shortcoming of common motion detection, a background-matching algorithm was proposed to solve such problem. When motion was detected, a Kalman predictor was used to estimate the object's position in the image. Combining increment PID controlling approach with dead zone, the camera can track the moving object steadily and reliably. The experiments proved the effectiveness of the algorithm.