Face Recognition System Based On Video Image Research And Implementation

With the continuous development of face recognition technology,its application potential and commercial value continue to receive attention and recognition from the society.However,in actual application scenarios,due to the diversity and complexity of the environment,as well as the multiple poses and angles of the face,the accuracy and timeliness of the face recognition system is greatly disturbed.In view of the above difficulties,this paper is committed to developing an efficient and stable face recognition system.The system is divided into two parts: face detection and face recognition.Based on the existing face detection and face recognition algorithms,this paper innovatively proposes a face detection algorithm combining skin color detection and improved Adaboost,and a 2DPCA face recognition algorithm based on global feature extraction of the face.The experimental results show that the system can quickly and accurately recognize faces while capturing video in real time.The main research contents of this article are as follows:(1)The development history and trends of face recognition technology in recent years are introduced,the face detection algorithms and the classification of face recognition algorithms are summarized,the commonly used face libraries are listed,the key problems in the current face recognition field are proposed.(2)The first core module of a face recognition system is a face detection preprocessing module.In order to achieve rapid separation and accurate localization of face targets from video,the face detection preprocessing module introduces moving object detection,skin color detection,and geometric feature models in detail,and makes a complete discussion of the principle and application process of each algorithm.(3)The second core module of the face recognition system is the face detection module.The face detection module focuses on the principle and implementation process of the Adaboost algorithm,and improves the traditional Adaboost algorithm from the perspective of Haar eigenvalues and weight updates.Combining the three pre-processing steps of motion analysis,skin color detection and geometric features,a face detection module experiment was designed to prove that the improved algorithm can reduce the face false detection rate by 17.2% and improve the face positive detection rate by 4.5%,and short the detection time to 25%.(4)The third core module of the face recognition system is the face recognition module.The face recognition module focuses on the PCA and 2DPCA algorithms based on global feature extraction,and compares the advantages and disadvantages of the two algorithms.Experiments show that 2DPCA is more suitable for face recognition feature extraction than PCA,which improves face recognition rate by about 5% and greatly stabilizes face recognition time.(5)Based on Visual Studio 2017 and Open CV development tools,a face recognition system based on video images was designed.Experiments show that the system has high face detection rate and face recognition rate,low false detection rate and short recognition time,which has good practical value.