The Impact of Changing Resolutions on Face Recognition

Many factors such as blurring effect, expression, illumination may affect the face recognition accuracy. In this paper, we focus our attention on the impact caused by resolution. In the video surveillance system, facial images in different resolutions may be detected, it is necessary for us to find out the impact of changing resolutions on different face recognition algorithms. A database has been collected of different facial images in different resolutions. Moreover, several experiments have been carried out to find the impact of changing resolutions on face recognition. We use the classical PCA, LDA and LBP method to extract the features and then use the nearest neighbor method to classify these facial images. Comparisons have been made among these classical algorithms in order to find out which method is the most effective in different resolutions. As a result, a confidence interval is obtained in which the algorithm performs much better than any other regions.