We proposed a novel face representation model, namely monogenic binary coding (MBC), based on the monogenic signal representation. One of the best merits of MBC is that it has much less time and space complexity than the widely used Gabor transformation based local feature extraction method. Through multi-scale monogenic signal representation, three kinds of features (e.g., local amplitude, local orientation and local phase) can be generated, and then encoded by the proposed monogenic local variation coding and monogenic imagery intensity coding procedures. The produced MBC pattern maps are used to compute the statistical features (e.g., histogram), which are then used to measure the similarity for face recognition. The extensive experiments on the benchmark face databases, including FERET, FRGC 2.0, Multi-PIE and PolyU-NIR, clearly showed that the proposed MBC methods not only have significantly lower time and space complexity than the state-of-the-art Gabor feature based face recognition methods, but also have very competitive or even better recognition rates.
Yang, Meng, Lei Zhang, Simon Chi-Keung Shiu, and David Zhang. “Monogenic binary coding: An efficient local feature extraction approach to face recognition.” IEEE Transactions on Information Forensics and Security 7, no. 6 (2012): 1738-1751.