Extended Discriminative Robust Local Binary Patterns Edge Detection Technique for Efficient Face Recognition based on Universal Gravity using Machine Learning

Abstract

In the face recognition system finding the exact shape or edge is very tough task in low contrasted and rotation variant faces to improve the accuracy rate of recognition. In this research a novel edge detection technique based on the universal law of gravitational force is proposed. The algorithm considers every pixel in the image to be a celestial body, each of those grayscale intensity corresponds to its mass. As a result, every celestial body exerts forces on its surrounding pixels and receives forces in return from them. The universal gravitational force is calculated by determining the direction of signal variations, magnitude and based on these Vector sums of all gravitational forces in both the horizontal and vertical directions are computed. As a result of their high gravitational force magnitude along a specific direction, edges can be identified in low contrasted and rotation variant faces. The method produced better recognition rate over the DRLBP technique because clear edges are extracted to identify the exact face features shapes.