In this paper a face-recognition algorithm with a confidence evaluation function for batch of SIFT feature is presented. Confidence evaluation function is rarely used in traditional face recognition, which is an important index in future recognition. In our face-recognition algorithm, two main steps are provided, that is primary election and strict identification. Adaboost algorithm can detect rough features to collect candidate face regions, it works as primary election algorithm. SIFT can describe the detail features in the face regions, the confidence evaluation function for batch of SIFT feature is highly distinctive, and it work as strict identification algorithm. This confidence evaluation function is a reliable measurement for matching multi-candidate regions containing invariant features. And, it can also be used in image retrieval, object recognition.

Date Published: 30 October 2009
PDF: 7 pages
Proc. SPIE 7495, MIPPR 2009: Automatic Target Recognition and Image Analysis, 74953O (30 October 2009); doi: 10.1117/12.832336

Published in SPIE Proceedings Vol. 7495:
MIPPR 2009: Automatic Target Recognition and Image Analysis
Tianxu Zhang; Bruce Hirsch; Zhiguo Cao; Hanqing Lu, Editor(s)