Survey of Occluded and Unoccluded Face Recognition

Computer Vision has been widely developed in recent years as the development of AI, as people want to let AI be able to see things. In all of the aspects of computer vision, face recognition seems to be one of the most important parts as for AI. People mainly want it to help them recognize the face which may be used for the security like the brush face payment. Also, as there are more and more criminals who know shelter their faces for not let the monitor to recognize their face and face recognition by the computer may solve this issue as there is a limitation for human to recognize the face and also is exhausting. In the last decade, face recognition mainly uses some linear or nonlinear models or neural network methods, which has a low accuracy and the database for these methods are small. Moreover, some of the methods may not be able to solve the picture which is occluded or is low-quality. This paper provides a recent research trends for face recognition in unobstructed, occluded and low-quality pictures. To make it easier to understand, tables of different applications, databases, methods are provided. This paper concludes by proposing some challenges may still involve in face recognition as well as some expectations for the future research.