Title: CONTENT BASED FACE IMAGE RETRIEVAL IN WALSH HADAMARD TRANSFORM (WHT) DOMAIN

Year of Publication: 2012
Page Numbers: 101-106
Authors: W.S. Besbas, M.A. Artemi, R.M. Suliman, M.M. AL_Rjebi
Conference Name: The International Conference on Computing, Networking and Digital Technologies (ICCNDT2012)
- Bahrain

Abstract:


This paper investigates the problem of features selection, for Content Based Face Image Retrieval in Walsh Hadamard Transform (WHT) domain. The face image features extraction in the spectrum domain of natural and sequency order Walsh Hadamard Transform is analyzed. The coefficients of WHT contain integer values only, and the Hadamard transform is a very fast transform, it can be implemented using additions and subtractions operations only. Sheffield face database is used to evaluate the performance of different methods for constructing the feature vector. Euclidian and City Block distance measures are analyzed during this evaluation. The investigation concludes that a proposed method of selecting small number of coefficients from first-row, first-column, and the diagonal, of the face image in WHT spectral domain, achieves high retrieval rate and less computational complexity for CBIR systems.