Face recognition algorithm based on stack denoising and self-encoding LBP

Abstract To optimize the weak robustness of traditional face recognition algorithms, classification accuracy rate is not high, operation speed slower, so a algorithm based on local binary pattern (LBP) and stacked autoencoder (AE) proposed. The advantage LBP texture structure feature image as initial sparse (SAE) learning, use unified mode operator to extract histogram blocked image, connect form features entire image. It used input AE, extraction done, realize images. Experimental results show that LBP-SAE Yale database has achieved 99.05%, it further shows higher than classic algorithm; strong light changes. Olivetti Research Laboratory library developed method more robust changes better effects compared algorithms standard stack AEs.