Fusing Frequency, Spatial and Color Features for Face Recognition

This paper presents a face recognition method by fusing the frequency, spatial and color features for improving the face recognition grand challenge performance. In particular, the frequency features are extracted from the magnitude, the real and imaginary parts in the frequency domain of an image; the spatial features are derived from two different scales of a face image; and the color features are from a new hybrid color space, namely, the RIQ color space. Specifically, every color component in the RIQ color space has two scales: Scale 1 image and Scale 2 image (with a larger face region). First, the frequency feature extraction procedure applies to all the Scale 1 and Scale 2 color component images. Then, an improved Fisher model extracts discriminating features from the frequency data for similarity computation using a cosine similarity measure. Finally, the similarity scores from the three component images in the RIQ color space are fused by means of a weighted summation at the decision level for the overall similarity computation. Experiments on the Face Recognition Grand Challenge (FRGC) version 2 Experiment 4 show that the proposed method achieves the face verification rate (corresponding to the ROC III curve) of 82.8% at the false accept rate of 0.1%.