Face and Expression Recognition Using Local Directional Number Pattern

Local directional number encodes the directional information of the faces textures in a dense way, producing a more discriminative code than current methods. We figure the structure of each micro-pattern with the support of a compass mask that extracts directional information, and encode such information using the well-known direction indices and sign. which allows us to distinguish among similar structural patterns that have different intensity transitions. Then we divide the face into several regions, and take out the distribution of the LDN features from them. Then, concatenate these features into a feature vector, and we use it as a face descriptor. In which descriptor performs consistently under illumination, noise, expression, and the time lapse variations.