Research on face recognition based on DRNLGBP

Abstract

For facial recognition of those wearing glasses, this paper first proposes a DRNLGBP algorithm based on DIMPCA and non-uniform local Gabor binary pattern method combined with random sampling. The algorithm firstly extracts and supplements the contour by the DIMPCA, and then it adopts the local Gabor binary pattern to represent the facial image, by which characteristic information of multi-scale and multi-direction can be extracted and can make the recognition less sensitive to illumination and more robust to noise. The paper also describes the non-uniform region division strategy, which can ensure sufficient image spatial information acquisition while treating the content information differently at the same time with weakening interference to adjustment from glasses. The experimental result shows that this method is more effective for facial recognition rate improvement for those wearing glasses than traditional methods, significantly reducing the influence of surrounding light degree.