Research On Object Tracking System In Intelligent Video Surveillance

In the present world, unsafe and unharmonious factors seriously threaten the safeties of country, society and people. Video surveillance system is one of the main technical means available to effectively prevent and eliminate hidden danger in fields of public security by real-time monitoring. However, traditional video surveillance system has some shortcomings and is unable to cope with complex scenes and behaviors. So it is imperative to develop intelligent video surveillance system (IVSS) which is based on the techniques of object tracking (behavior recognition). The object tracking system is the key techniques in intelligent video surveillance, including moving object detection and object tracking. The main attention of this paper is the real-time target tracking in the complex background; we expect to improve the robustness and stability of object tracking by fusing the multiple cues. The main contributions of this thesis are summarized as:A Bayesian tracking algorithm based on the covariance region descriptor is proposed. The covariance descriptor can fuse different low level visual features and the proposed algorithm encodes it in the Bayesian tracking framework to handle complex background. Moreover, the fast covariance computation is extended to Bayesian tracking framework, which makes the tracking process more efficient.This paper presents a novel tracking algorithm that fuses multiple features based on feature uncertainty measurement. It is based on the fact that tracking failure of particle filter often happens in the cases of low discriminative abilities of the observed features and disperses distributions of the sampled particles. To handle this failure, we define a new feature uncertainty measurement method to adaptively adjust the relative contributions of di?erent cues. Then we introduce a self-adaptive cue fusion strategy to overcome the shortcomings of product and sum fusion ones. This strategy e?ectively sharpens the distribution of the fused posterior, and makes the tracking less sensitive to noises. Thereby, the tracking robustness is improved. An extensive number of comparative experiments show that the proposed algorithm is more stable and robust than the single feature, product fusion, and sum fusion tracking algorithms.