Analyzing the performance of Viola Jones Face Detector on the LDHF database

摘要

Face Recognition at night time at a large standoff is a very important task in surveillance especially in defense applications. Face Detection being the primary step in Face recognition thus results to be a very challenging task when used on night time images taken at a long distances. This paper analyses the performance of the Viola Jones Face Detector which is termed to be a benchmarked among all available face detectors on the LDHF database. The LDHF database contains both visual and NIR images taken at various distances such as 1m, 60m, 100m and 150m. This work considers only the NIR images as enough work is done on visual images. The outcome shows that the detection rate decreases as the standoff distance increases. Also it brings out the need for improving the performance of the Viola Jones Face Detector on night time images.