Recognition of blurred faces using local phase pattern

A novel local phase pattern (LPP) descriptor is proposed for blurred face recognition. The descriptor extracts blur invariant features - the local phase pattern information of the Fourier transform computed locally in a window for every image position. In face image analysis, histograms of LPP labels computed within local regions are used as a face descriptor similarly to the widely used local binary pattern (LBP) methodology for face image description. The LPP descriptors are assessed in comparison with LBP and local phase quantisation. The results show that LPP achieves a promising recognition rate, even when the amount of blur significantly increases.