| dc.description.abstract | People tracking and behavior recognition has been emerged to play critical roles in numerous areas including entertainment, robotics, surveillance, etc. In order to make an approach of people tracking and behavior recognition to be widely used, the convenience to users, the simplicity in installation, and the reasonable prices for equipment are the main factors to be considered. The conventional work of capturing human motion is wearing sensors, however, the extra burdens of wearing sensors all of the time and sensors could go unworn, making the task unreliable.
Tracking and recognizing human behavior from images obtained by a monocular camera may be an option. However when taking a 2-D picture of a scene with a monocular camera, the appearance of a person in a 2-D image might pose many possible configurations in 3-D, the depth information will be loose and results could be affect by the lighting conditions. In this dissertation, another solution is concerned with the uses of multiple depth camera to overcome the limitations of the monocular image-based approach, but occlusions still reduce such methods’ accuracy.
To address these problems, we propose a pairwise trajectory matching scheme from multiple cameras for people tracking, using curve clustering to fuse the tracking trajectories. For behavior recognition, a time-variant skeleton vector projection scheme using multiple infrared-based depth cameras is developed by combined proposed time-variant basis vector and proposed occlusion-based weighting element generation. The experiment results shows the proposed method achieves less tracking distortion, superior behavior recognition accuracy and involves less computational complexity compared with other state-of-the-art methods for practical testing environments. | en_US |