NABILD: Noise And Blur Invariant Local Descriptor for Face Recognition

Abstract Most of the existing local descriptors unable to perform well in front noise and blur variations. Additionally there are very less persist literature which invariant. To remedy this challenge proposed work launch novel descriptor under changes so-called Noise Blur Invariant Local Descriptor (NABILD). With respect two artificial noises i.e. Gaussian White (GWN) Salt & Pepper (SPN) with image blurring, NABILD is introduced. Precisely takes essentials performed The first one Median Robust Extended LBP based on Neighborhood Intensity (MRELBP-NI) second Multiscale Phase Quantization (MLPQ). MRELBP-NI effective controlling GWN SPN due capturing microstructure macrostructure information. LPQ efficient invariant as it quantizes phase. By considering merits both these their features integrated into framework called NABILD. lower down feature dimension FLDA deployed classification conducted by SVMs. Experiments ORL face dataset confirm strength against other tested Various methods also outclassed