Face Detection and Authentication in a Large Space

Abstract

Moving pedestrians are too small to grab the clear and recognizable face images in current surveillance systems. In this study, a face detection/indexing scheme has been developed using a dual-camera based capturing module in a large open space. Two cameras, a 'sense' camera and an 'object' one, are calibrated and cooperated. This device not only tracks multiple targets but also gets the close-up snapshot images. Face images with clear and high quality are grabbed and collected at the enrollment stage. A personal feature space and a threshold are constructed for effectively verifying the legality of an entrant. Some experimental results are given to show the validity of the proposed approach.