Computer Vision by Tracking and Recognition

尽管传统的计算机视觉系统已经逐帧解决了许多识别问题，但未来仍需要更高效的视觉系统来开发可能的应用。本研究提出了一种新的计算机视觉方法，通过引入一个统一的跟踪和识别框架。在我们的方法中，在基于高效目标跟踪技术的统一框架中讨论了三个典型的识别问题。最后，我们通过在统一系统中创建包括面部运动跟踪，面部识别和表情识别在内的人机界面来证明我们方法的有效性。

Although many recognition problems have been solved frame by frame in the conventional computer vision systems, more efficient and effective vision systems are desired for developing possible applications in the future. A novel approach to computer vision is proposed in this research by introducing a unified framework of tracking and recognition. In our approach, three typical recognition problems have been discussed in the unified framework based on an efficient object tracking technique. Finally, we have shown the effectiveness of our approach by creating a human interface including the face motion tracking, face recognition and expression recognition in the unified system.