Abstract:

Currently the intelligent monitoring system with real-time detection, tracking and analysis is the future trends. In this paper, we propose an algorithm to detect and track a moving object in real time surveillance environment. Firstly we adopt background difference method to detect the moving object.Then we use a local strategy to update changing background to adapt the complex environment,such as light variant,mutual transformation of foreground and background.In the tracking process,keypoints are obtained in different scale space to enhance the robustness of algorithm in object occlusion.Lastly mean shift method is used to estimate the position in next frame of object.The experiment result demonstrate s the feasibility and validity of our algorithm.

Key words: object detection, tracking, scale space, mean shift