Local symmetrical patterns-based feature extraction model (LSP-FEM) for efficient face recognition

In the applications of computer vision and pattern recognition, facial image processing has been a great issue to focus on for providing efficient solutions for face recognition. General face recognition models can be classified into two types, geometry-based and appearance-based feature models, which deal with global feature data and facial textures respectively. Normally the performance of an adaptive face detection model increases with an increase in the number of training images. In this study, a novel model called Local Symmetrical Patterns based feature extraction model (LSP-FEM) for efficient face recognition was developed. The model incorporates Local Symmetrical Patterns (LSP) to recognize the input human facial samples. Moreover, the proposed LSP-FEM computes the symmetry of each pixel in all eight directions of facial images. For an efficient recognition process, a facial image is considered as a collection of LSP codes. Furthermore, the experimentation was carried out using benchmark datasets called the FERET dataset, Extended Yale-B dataset and Olivetti Research Laboratory (ORL) dataset images. The results show that the accuracy rate of face recognition is higher than that of the existing models.