الفهرس | Only 14 pages are availabe for public view |
Abstract This thesis addresses the problem of human activity recognition in realistic video data, such as movies and online videos. Automatic and accurate recognition of human activities in video is a fascinating capability. The potential applications range from surveillance and robotics to medical diagnosis, content-based video retrieval, and in- telligent human computer interfaces. The task is highly challenging due to the large variations in person appearances, dynamic backgrounds, view-point changes, lighting conditions, action styles and other factors. Another goal of the research presented in this doctoral thesis is to explore and establish theories and methodologies for accurate representation and recognition of human activities in videos. For the methodological contributions of this thesis, multiple approaches involving diverse conceptualizations are developed to represent and recognize human activities from video sequences. Moreover, we investigate global representation features based on spatiotemporal orientation energy and template matching for the representation and recognition of human activities. For our {uFB01}rst approach, we present a new approach for human activ- ity recognition based on global features and different feature selection techniques. The main contribution of this approach is twofold. First, reliable global features as video representations are employed for the task of activity classi{uFB01}cation. |