Palmprint recognition using local and global features

In this paper we propose palmprint recognition technique by using a fusion of local and global information extracted from palmprint image. Local information are extracted using discrete cosine transform computed in several image region while global feature are extracted using LDA. The matching score of both features are fused using sum rule to produce a fused score. The proposed method is able to increase discrimination power and preserve low frequency coefficients which are important for recognition and verification process. The proposed method is tested using PolyU dataset and the highest recognition rates is 97%. The best verification rates is Genuine Acceptance Rate (GAR) = 98% at False Acceptance Rate (FAR) = 0.1%.