Research And Implementation Of Face Detection Algorithm In Video Image

Since the 21st century,science and technology in the Internet field have been constantly updated and iterated.Biometrics technology is even more brilliant in today's artificial intelligence era.As a branch of biometrics,face recognition technology has gradually become a hot research topic..Face recognition technology includes four parts:face detection,image preprocessing,feature extraction and matching recognition.Face detection is the first step in face recognition technology.Face detection is a computer technology.It refers to a certain image or video,using a certain strategy to detemiine whether it contains a human face,and accurately mark the face,position.Early face detection mainly focused on detecting the position of a face in a static picture of a single background After years of development,the face detection technology is now more used for dynamic video detection in complex backgrounds.Because the face detection results are easily interfered by environmental factors such as obstructions,angles,and light,the commonly used face detection algorithms generally have the disadvantages of low detection rate,high false positive rate,and inability to detect multi-angle faces.In this paper,the face detection algorithm in video images is studied.The related techniques are used to process the frame images and then face detection,which effectively solves the defects of traditional algorithms.The main work of this paper is as follows:1?The research background of face detection and the practical significance of face detection algorithm in video images are introduced in detail.The classification of four face detection methods is summarized,and the perfonmance evaluation index of face detection algorithm is introduced.2?The moving target detection technology,image deaoising technology and skin color detection technology used in this paper are introduced in detail,including the algorithm ideas based on various technologies and the implementation process of the algorithm;based on the Viola-Jones real-time face detection algorithm Three new Haar-like features are used to train strong classifiers,and the resulting cascade classifier can be adapted to multi-pose face detection.3?A real-time face detection system in video images is designed and implemented on the MATLAB platform.Firstly,the moving target is extracted by using the motion information in the video,then the frame image is denoised and the skin color is detected,and the frame image is classified into two types of skin color pixels and non-skin color regions,all of which are in the video frame image of the cascade classifier.Before the sub-window is detected,the decision fimction is used to determine the proportion of the skin color pixels in the sub-window.This step effectively rejects a large number of sub-windows that do not contain human faces,and improves the algorithm-to-multi-pose while ensuring high detection speed.The detection rate of the face.