Research on Face Recognition Technology Based on Deep Learning

The traditional face recognition technology requires a lot of computation time for human face feature extraction, and the recognition accuracy is limited by the extracted face features. Aiming at the above problems, this paper studies face recognition technology based on deep learning. For the image to be recognized, the HOG feature is used to detect the face target in the image. In order to avoid the influence of facial expression and posture, face alignment processing is performed. After extracting the face recognition features in the image, the neural network is used to realize face recognition. Through comparison experiments with traditional face recognition technology, it is verified that the recognition technology using deep learning improves the recognition efficiency by about 1.5 times and has better recognition adaptability.

I Introduction

Face recognition is a complex classification problem. Face recognition is a supervised classification problem. It refers to training samples after labeling the personal identity information of the face image in advance, and then classifying the face image to be recognized into a specific personal identity information set with a label. Face recognition technology is one of the important technologies of information security, computer network information verification and e-commerce verification. With the rapid development of the Internet, many biological technologies have been applied in the process of information security, computer network login and e-commerce. Among them, face recognition is one of the important parts in the field of biometrics, and plays an irreplaceable role in information security. With the continuous development of camera phone products in recent years, the way to obtain facial images has become more and more simple. Face recognition is developed from image recognition, and the recognition effects of different algorithms vary greatly [1].