Research on Face Recognition System Based on Deep Convolutional Machine Learning Model

Abstract

This paper proposes a face recognition system based on a deep convolutional neural network algorithm. Firstly, according to the distribution law of pose face, the nonlinear manifold space of pose face is divided into different manifold layers and local subspaces. At the same time, this paper defines the low-level feature construction method for the pose face in the local subspace to realize the face sample expansion with pose change. Then this paper obtains a self-learning deep convolutional neural network through network structure initialization, global and local adaptive expansion of the network structure. In this way, the deep nonlinear feature extraction and recognition of pose-changing faces is realized. The experimental simulation shows that the recognition accuracy rate of the algorithm on Clubfeet, AR and ORL face databases reaches 98.89%, 99.67% and 100% respectively. The algorithm has a fast convergence rate.