Research And Implementation Of Face Recognition System Based On Video

Face recognition technology is a safe and convenient biometric technology,which has been a hot topic for domestic and foreign scholars.With the continuous development of computer vision and image processing technology,face recognition technology has become increasingly mature,and its efficiency and accuracy have been greatly improved.Face recognition technology is also more diversified and has been widely used in human-computer interaction,video conferencing,access control systems,and smart homes.Face recognition technology provides a lot of convenience for people's daily life and has great practical value.Face recognition is currently mainly used for identity verification.Identity verification is a very common part of our daily life.The company needs to perform authentication for card punching,station ticket checking,security systems,etc.In these scenarios,the general traffic is large.The method of requiring authentication takes speed and accuracy into consideration.Face recognition is very suitable for identity authentication in these large-scale scenarios,which can ensure accuracy and meet speed requirements and save unnecessary manpower and material resources.Due to the diversity and complexity of the actual use scene environment factors,the accuracy and stability of the face recognition system will inevitably be interfered.Therefore,it is also difficult to design a face recognition system with high accuracy and robustness in specific scenarios.Based on the analysis of the related techniques of opposite face recognition,this paper divides the system into two parts,face detection and face recognition.Then,using the relevant face recognition related algorithm,through algorithm improvement and data set experiment,a face recognition system based on real-time video is designed and implemented.The face recognition system can collect video in real time and quickly perform face recognition,and has achieved good experimental results.This paper first studies the face detection related algorithm,and mainly studies the face detection algorithm of Adaboost based on Haar feature.After the introduction of the principle,based on Open CV using the extended face dataset,a sample with a tilt angle is added to the face sample to train the multi-pose Haar-based Adaboost face detector.The experimental results show that the classifier has a good recognition rate for people who can quickly recognize faces and have a certain tilt angle,which proves the efficiency and accuracy of detecting faces based on Adaboost face detector.The face detector based on Haar feature is selected by contrast experiment.For the problem of high false detection rate,the secondary screening process based on skin color model is proposed,which significantly reduces the false detection rate of face detection.After obtaining the face image,perform some pre-processing and save to the face database.Then the face recognition function is implemented,the uniform LBP features of the face image are extracted,and then the SVM classifier is trained to realize face recognition.The experimental results show that the LBP feature and SVM classification algorithm can improve the training speed and recognition speed of the model,and have a high accuracy.This paper introduces the related techniques and algorithms of face recognition and analyzes various methods of face detection and face recognition in detail.The face detection method for face verification of face skin model is proposed,and then LBP and SVM are combined to realize people.Face recognition function,finally based on Open CV platform and Qt development framework design to achieve a full-featured video-based face recognition system,which has both real-time and accuracy and has certain practical value.