Combining face detection and people tracking in video sequences
Etienne Corvée, François Brémond · 2009
Face detection algorithms are widely used in computer vision as they provide fast and reliable results depending on the application domain. A multi view approach is here presented to detect frontal and profile pose of people face using Histogram of Oriented Gradients, i.e. HOG, features. A K-mean clustering technique is used in a cascade of HOG feature classifiers to detect faces. The evaluation of the algorithm shows similar performance in terms of detection rate as state of the art algorithms. Moreover, unlike state of the art algorithms, our system can be quickly trained before detection is possible. Performance is considerably increased in terms of lower computational cost and lower false detection rate when combined with motion constraint given by moving objects in video sequences. The detected HOG features are integrated within a tracking framework and allow reliable face tracking results in several tested surveillance video sequences. These trajectories show unidirectional patterns of people trajectory and occur in restricted scenarios such as in corridors. However, while a person moves across the camera field of view, both the person frontal and profile view of his face are likely to be captured. The proposed face detection algorithm learns object features using a generic approach. Objects are represented by feature vectors in section 3 and a description on how different poses of facial features can be learned using a K-mean algorithm is given in section 4. Due to the large complexity occurring in grey level distribution of people face, a cascade of classifiers is designed in section 5 to speed up the detection process. In the proposed approach, the fact that most surveillance cameras are static and viewing static background scenes allows relatively fast segmentation of 2D moving objects where faces are assumed to occur, hence restricting the face searching process. Faces are tracked by integrating the detected features within a tracking framework presented in section 6. Results of detected and tracked faces are shown in section 7. 1