Online Video Face Recognition System Based On LBP And Ridge Regression

With the development of the intellectualization and information technology, thehuman-computer interaction becomes the most effective communication methodbetween human and intelligent computer. In this situation, the facial recognitiontechnology has been widely researched in the field of computer, electronics, imageprocessing, automation, and pattern recognition and so on.At the same time it hasimportant application value in the air, criminal investigation, entrance guard, the entryand exit, and other fields. The goal of this thesis is to complete the face recognitionsystems in online video, and to study face recognition principle, method and relatedtechnology. Face recognition system includes face detection module, feature extractionmodule, face decision module. The main work of this thesis is as follows:（1） The face detection. Face detection is an important problem of facerecognition system. According the analysis and research of face detection algorithm,this thesis researches face detection AdaBoost algorithm first, and then realizes thereal-time face detection of the captured video streamingby the USB camera by themethod of detecting face before locating eyes.（2）Feature extraction. In recent years, the use of Local Binary Pattern （LocalBinary Pattern, LBP）texture feature has achieved some success in face recognition.Based on the analysis of original LBP operator, equivalent model LBP operator, thisthesis uses the equivalent model of LBP operator for feature extraction, and combinesPCA method to reduce the dimension for high-dimensional data.（3） Determine face. After the face image feature extraction, the choice ofclassifier is very important. By studying the Extreme Learning Machine（ELM）,Support Vector Machine（SVM）, regression method of facial recognition and ridgeregression in Extreme Learning Machine,this thesis proposed a method of combinationof LBP and ridge regression algorithm for face recognition. The design of classifier iscompleted through the training of characteristics to get the weights of the output datamatrix. We can use classifier to classify the images which have not been identified andget the recognized faces.Based on the Visual Studio2010C++development environment and the libraryof OpenCV, the software to write to realize computer vision algorithm. It can accomplish the real-time face detection through USB camera to get a face database fortraining, finally realizes the function of face recognition.