Summary: In this paper, we propose a novel color face feature extraction approach named statistically orthogonal analysis (SOA). It in turn calculates the projection transforms of the red, green and blue color component image sets by using the Fisher criterion, and simultaneously makes the obtained transforms mutually statistically orthogonal. SOA can enhance the complementation and remove the correlation between discriminant features separately extracted from three color component image sets. Experimental results on the AR and FRGC version 2 color face image databases demonstrate that SOA achieves better recognition results than several related color face recognition methods.