Research On Face Detection Method Under Complex Illumination

Face detection is widely used in people's life and learning.Especially in recent years,due to the rapid development of deep learning,the research of face detection has reached a very deep level,and it is developing towards more complicated environments,such as complex illumination,facial occlusion,half-face detection and so on.Among them,face detection based on complex illumination is the most difficult point in current research,and it is also the main research direction in the future,the accuracy of face detection under complex illumination conditions still needs to be improved.In order to improve the detection accuracy of face detection algorithm under complex illumination,this paper firstly analyzes the existing face detection methods in detail,and then the data set,data enhancement technology and detection algorithm are improved to improve the average detection accuracy of face detection algorithm under complex illumination.The specific work is as follows:Firstly,we build a face data set under complex illumination,because the existing data sets are less concerned with complex illumination,this paper uses two sources to construct the data sets under complex illumination,one is to screen the existing data sets,and the other is to screen the images collected by the image acquisition device.Due to the large number of images that need to be filtered,an image illumination complexity filtering algorithm is proposed in this paper to complete the screening task,and then the face data set under complex illumination is constructed.Then,we study data enhancement technology under complex illumination.The current data enhancement technology is based on the enhancement under normal illumination conditions and it does not take into account the complex illumination conditions.Therefore,in this paper,the data enhancement technology under ordinary illumination is studied and analyzed in detail,and in order to improve the face detection accuracy of trained network model under complex illumination,the data enhancement technology under complex illumination is used to process the data set,which expands the number of data set and adds a variety of complex illumination conditions.Finally,we propose a face detection algorithm for complex illumination in this paper.The existing detection algorithms are analyzed in detail,and the network structure and the processing methods of detection stage are improved properly.Then,a face detection algorithm for complex illumination is proposed.Through related experiments,it is proved that the complex illumination face data set constructed and the complex illumination data enhancement technology in this paper can improve the face detection accuracy of the network model under complex illumination conditions.The face detection algorithm for complex illumination proposed in the paper can achieve higher detection accuracy.