Research And Verification Of The Face Detection Algorithm

Face detection means determining whether there is human face in the input image,if existing,it will mark the face area,and further determine the location and size of the face.As the first step of face recognition,face detection has attracted great attention of researchers.With the continuous progress of science and technology and the improvement of people's security awareness,face recognition technology has been applied to more and more fields.For example,the missing children mutual-assistance system,the public security video surveillance system,the sign-in system and so on.It can be seen that the research of the face detection algorithm will become a hot topic.This paper improves and designs the face detection algorithm based on skin color model,template matching and Gentle Adaboost and validates them.The specific work can be roughly expressed as follows:1?Research and Application of Face Detection Method Based on Skin Color Model.As we all know,skin tone is an important information for human face,which will be a great help for face detection in color images.Any details of the human face does not affect the detection results only from the skin tone.It greatly reduces the detection time and the amount of calculation.We have completed the detection of human faces in a frame of video through the construction of the skin color model.The method of light compensation is used before the face detection to reduce the influence of illumination on face detection.Then,the captured image is transformed into the YCbCr color space with good color clustering characteristics.Finally,the skin color model is created by the elliptical skin color model in the YCbCr color space,and the traditional morphological processing is improved into the morphological processing proposed in this paper.Finally,the face area is identified.2?Multi-angle Face Detection Based on Improved Template Matching.In order to solve the problem of template matching in face detection for a long calculation time and low face detection rate in different angles,this paper proposes a multi-angle face detection algorithm based on Gaussian mixture model and improved template matching.The necessary light compensation processing is performed first.Then the skin color model is created by the Gaussian mixture model in the YCbCr color space,and the candidate regions of the face are obtained through appropriate moiphological processing.Next,average lace lcmplates of seven diflercnt angles are created and calculated,and formed into a face template database.The template with the closest angle is selected according to the angle of the face to be detected,so that the location of face can be accurately marked.Experiments show that the algorithm is 8%higher than the traditional template matching algorithm in the detection of different numbers of multi-angle faces and reduces the detection time.3?A Face Detection Algorithm Based on Multi-feature Fusion.There are few types of Haar-like rectangle features,which leads to the problem that the classifier training time is too long due to the large number of feature quantities required in the description of the face.Local Binary Patterns(LBP)are used in this paper.Considering the inadequacy of basic LBP features in face detection,unified MB-LBP features and unified rotation-invariant LBP features are used to describe local texture features of faces.Considering the disadvantage of MB-LBP features and rotation-invariant LBP features on face edge information,the edge azimuth field features based on Canny operator are combined with the above two features to describe face information.Finally,the Gentle Adaboost classifier is designed to classify all the extracted features.The three features can not only describe the local face information but also the whole face information,which greatly improves the detection rate and detection speed of the face with multiple poses and different rotation modes.The Gentle Adaboost classifier has superior performance compared to other Adaboost classifiers when applied to face detection under the same conditions.