REAL-TIME ALGORITHM of OBJECT TRACKING with ERROR HANDLING

Abstract

Video surveillance applications have to support objects authentic in case of multiple objects tracking. During tracking process errors, caused by different circumstances, such as objects intersections among themselves, intersections with objects on the foreground, objects shape changes are possible. This article describes multiple objects tracking algorithm, based on system of object characteristics. The probabilistic model is based on the Kalman filter. The system is fully automated and does not require any manual input during initialization or execution.

Keywords

• трекинг множества объектов
• модель движения
• фильтр калмана
• окклюзия