Object tracking, an intermediate-level vision part, plays an important role in visual analysis and understanding of object motion. The goal of object tracking is to detect, localize and track moving objects in videos. By visual tracking, we obtain the motion parameters, the poses and the trajectories of objects, which lays a solid foundation for the high-level activity understanding and recognition. Although object tracking is really a hot research topic in recent years, there do exist many theoretical and technical problems, especially in the cases of noise, motion blurring, illumination changes,occlusion etc. Trajectory pattern learning is part of high-level object motion analysis.It is simply viewed as an unsupervised trajectory learning problem. After trajectory learning, several trajectory patterns are obtained. For these trajectory patterns, pattern-specific retrieval models are learned. Subsequently, trajectory pattern retrieval can be completed by matching a query trajectory with the learned pattern-specific retrieval models. Therefore, the core of trajectory pattern learning and retrieval is how to learn and model trajectory patterns in an effective and efficient way. This thesis mainly focuses on robust object tracking and trajectory pattern learning. Specifically, we mainly discuss two topics: (1)appearance modeling based object tracking; (2)object trajectory pattern learning and retrieval. Experimental results have demonstrated the effectiveness and robustness of our methods. The main contributions of this thesis are summarized as follows: 1) A novel object tracking framework based on incremental tensor subspace learning is proposed. The framework online learns a tensor-based eigenspace appearance model by incremental tensor subspace learning. The appearance model not only reduces the spatio-temporal redundancy information of object appearance, but also fuses the spatio-temporal information from different views. As a result, the appearance model is robust to many complex conditions such as noise, motion blurring, illumination changes etc. Furthermore, we present a foreground segmentation framework based on incremental tensor subspace learning. The framework online constructs a low-dimensional tensor-based eigenspace background model. It is capable of effectively capturing the spatio-temporal distribution information of a scene, resulting in robust foreground segmentation results. 2) We present two novel object appearance models based on i...
修改评论