An Improved Hybrid Fast Face Recognition Algorithm

In recent years,research on object recognition,shape matching and pattern recognition in the field of computer vision has been paid more and more attention.It aims to develop a highly accurate facial recognition system while taking into account factors such as robustness and system uptime.Face recognition is one of the important topics in object discrimination and digital vision research.This system requires real-time reliable stability.Although the traditional face recognition system can achieve higher accuracy,its computational cost is also very high.The ability to correctly identify and classify facial images is determined by many variables,the most important being feature extraction and classification algorithms.This thesis mainly studies the feature extraction and recognition of face and face based on the combination of the whole method and the local method.The particle swarm optimization algorithm is used to optimize the classifier and feature selection to improve the accuracy of face recognition.This paper first discusses the research background and significance of face recognition,and briefly introduces the bottleneck and future development trend of face recognition technology.Secondly,the key problem of face recognition,namely the preprocessing method,is summarized,and then the feature extraction of the image is carried out.This work mainly uses the improved principal component analysis and simplified local pattern method to extract the features of the preprocessed image.Due to the high dimensionality of the face image and the large number of samples,the time cost of feature extraction increases,so we use wavelet transform to achieve fast dimensionality reduction.In the feature selection,the particle swarm optimization algorithm is mainly used.The algorithm has the characteristics of high stability and strong global searchability,which can neutralize the shortcomings of the classifier and help the learning ability and convergence speed.The highlight of this paper is the ability to perform high-precision face recognition with low computational cost.Finally,we use ORL,Yale and CK facial database for data analysis.The results show that our proposed method is slightly superior to the conventional methods of face recognition and the popular neural network and strip block method.