Abstract
This paper proposes face recognition based on dominant frequency feature and multiresolution metrics. Dominant frequency feature is a small part of face frequency component that represents face information. Dominant frequency feature is extracted by selecting small percentage discrete cosine transforms coefficients that have big magnitude value. Matching process is done by multiresolution metrics which calculate the level of similarity (score) between query and target face feature. The smallest score is concluded as the best likeness. The aim of the proposed methods is to provide face recognition that has good performance, need little training and query time, and require small feature size. The test carried out on four multi pose databases that have difference characteristics. The system shows good performance when compare to other approach.
Index
Terms
Available to subscribers and IEEE members.
References
Available to subscribers and IEEE members.
Citing Documents
Available to subscribers and IEEE members.
Cart
