Multi-level Fisher vector aggregated completed local fractional order derivative feature vector for face recognition

In this paper, we propose an image feature extraction method, multi-level Fisher vector aggregated completed local fractional order derivative feature vector (mFVFD), for face recognition. The novelties of our method are summarized as follows: (1) We propose multi-value local fractional order derivative feature vector to analyze the image local intrinsic edge information. (2) We define local multi-structure model to describe the variety and complexity of image structure feature. (3) We developed multiple kernel learning based multi-level Fisher vectors with different number of Gaussian components feature fusion method to capture different levels of the image characteristics. Extensive experiments are conducted on four standard face databases and the results have demonstrated that our proposed method outperforms the state-of-the-art.