Face recognition system

Abstract

The aim of this project is to develop a face recognition system based on Scale-Invariant Feature Transform (SIFT), which could extract distinctive features. The features generated by SIFT are highly invariant to image scaling and rotation, and partially invariant to change in illumination and 3D camera viewpoint. In order to achieve fast speed and better accuracy, face region is detected using “boosted cascade of simple features approach”. Image processing step then converts this region to Portable Gray Map (PGM) file. SIFT could extract features from the PGM file and finally a simple matching step is performed to match the extracted face with the target face. The output of the system would be matched percentage of the two faces. In this report, basics about face detection and SIFT would be studied first. Then four stages of the system would be covered in details, which are face detection, image processing, SIFT and matching. The results in each stag are also presented. Finally, processing time is roughly estimated to discuss the feasibility of applying this system to surveillance and Video Google application.