An Efficient and High Performance Feature Extraction Approach to Face Recognition Using Monogenic Binary Coding

Gabor features encoded by local binary pattern, could achieve state-of-the-art FR results in large-scale face databases.However, the time and space complexityof Gabor transformation are too high for many practicalFR applications. We propose a new and efficient local feature extraction scheme, namely MONOGENIC BINARY CODING (MBC), for face representation and recognition. The original signal is decomposed into three complementary components: amplitude, orientation, and phase in the Monogenic signal representation. Firstly we encode the monogenic variation in each local region and monogenic feature in each pixel, and then calculate the statistical features of the extracted local features. This is called local statistical feature extraction for face recognition (LSFFR). For the second phase of LSF-FR, many feature combination methods have been proposed. To exploit the discrimination information embedded in the amplitude, phase and orientation components of monogenic signal representation, in this paper we introduce an efficient and effective LSF-FR scheme, namely monogenic binary coding (MBC), which encodes the local pattern in different monogenic feature maps. The most commonly used strategy for discrimination is weighing the histogram feature extracted in different blocks. The block-based Fisher linear discriminant (BFLD) method is proposed to extract the low-dimensional discriminative features. For effective FR the local statistical features mined from the monogenic components (i.e., amplitude, orientation, and phase) are then fused for face classification.