Abstract:

This paper presents a detection and tracking algorithm for moving object in video sequence.A combination of the difference and the background subtraction based on codebooks background modeling,the algorithm can have accurate object detection and extraction of the moving object.It can also extract background with the existence of moving foregrounds.A kalman filter is used to predict the moving object's position at the next moment,so that by Camshift tracking algorithm,search and object matching can be made over a smaller range,thus reducing the amount of calculation,saving the searching and matching time and improving the tracking efficiency.Experiment shows that this algorithm is of practical value.

Key words: codebooks;background modeling;Kalman filter;Camshift