An improved low-resolution face tracker system using gradient logarithm field feature space and bi-cubic interpolation

Face tracking has become an active area of research it is very much helpful in applications such as video surveillance. The factors like illumination discrepancy, face pose variations, occlusions, and background clutter makes face tracking a challenging issue. Even though, a number of related work have been proposed to assist face tracking in unrestrained videos, the existing approaches may not be productive in video surveillance applications. This is mainly because of the fact that low-resolution face comprises of lesser informative data, and main transformations in illumination on the face make the tracking unproductive. In order to overcome this issue, this paper proposes a novel algorithm for illumination normalization and it is combined with the GLF for face tracking for improving the overall performance of the tracker system. Experimental results show that the proposed GLF-based tracker works well under significant clarification changes and outperforms many existing tracking algorithms.