Article Preview
Top1. Introduction
With the continuous growth of Internet information, there is a large amount of data on the Internet nowadays, making information security management extremely important. Therefore, the relevant departments have also strengthened their emphasis on information security. In the current situation, the most important personal information is identifying a person (Bong K, et al., 2017). Because of their reliability, universality, simplicity, stability and other advantages, biometric features have attracted more and more attention, and become a research hotspot incomputer vision and pattern recognition (Li Y, et al., 2018). Biometrics technology is to identify people by using the uniqueness and uniqueness of their biological features, including palmprint, iris, fingerprint, facial information, etc (Bong K, et al., 2017). Compared with fingerprint and iris information which are difficult to extract, face features are easy to obtain, easy to capture, easy to handle and non-contact. They are widely used as personal identity authentication information. Face detection is a biometric technology that automatically carries out facial feature information. It integrates digital image processing, pattern recognition and other technologies, and collects images or video streams containing faces by cameras or cameras for automatic detection and tracking (Long B, Yu K, Qin J, 2018).
Face detection refers to the use of face detection technology to identify people's identity information in images and mark it out. Face verification refers to using face detection technology to determine whether the current person to be identified is the same as the preset person (Yin X & Liu X, 2018). When traditional machine learning methods deal with raw data, they need to use a feature selection algorithm to transform the raw data into a more discriminating feature representation. In contrast, deep learning is an end-to-end learning method, which does not need to design features manually, and its learning process is simple, greatly reducing the time spent on feature selection (Deng J, 2017). Compared with the features extracted by traditional image description operators, the features extracted by deep learning are deep, abstract and complex. This in-depth feature can describe all aspects of human face well, and has strong robustness to the interference factors of environment and human face itself (Xiao Y & Xie X, 2019). Based on the idea of local features and deep learning, this article uses CNN to optimize face detection technology based on existing face local feature selection algorithms. It explores the face detection algorithm with better performance under the interference of light, expression and other self or external factors.
Face detection technology is to solve the problem of face image feature extraction. Like other image recognition problems, studying face detection is not only to solve the face detection problem but also to solve the significance of further target recognition. The face detection system has no high requirements for the performance of image acquisition devices. Common image acquisition devices such as mobile phones and cameras can be used for face detection, and other devices are not needed for assistance (Zhang B, 2019). Face detection is mainly based on the fact that human face features are quite different among different individuals, and it is a relatively stable measure for the same person. Because of the complexity of face changes, there are many difficulties in feature expression and feature extraction. Face detection has been widely used in human-computer interaction, entertainment, information security, video surveillance, medicine and finance. In the research of face detection optimization, the main contributions of this article are as follows: