One sample per person facial recognition with local binary patterns and image sub-grids

Local binary patterns (LBPs) are very commonly used to determine if a face in an image is a known person. However, accuracy is generally proportional to the number of training samples collected. The “single sample per person” (SSPP) problem focuses on identifying a person using only one training sample. Facial recognition from a single sample reduces the labor required to gather training data and enables some applications where only a single sample will be available. In this paper, we propose a method of improving the accuracy or efficiency of LBP-based face recognition by using a tree-based data structure to create “sub-grids” allowing for novel division patterns to be used in facial recognition, as opposed to the uniform grids used for most LBP face recognition. This method is then applied to the one sample per person problem where some patterns were found to require fewer regions within the image for comparable results to uniform grids.