Research On Adaboost Algorithm Optimization And Face Detection Recognition System

With the era progress and science & technology development,the face detection and recognition technology have gradually become the significant research topic in machine learning and model recognition scientific research field,which is of the highest academic research value and commercial application value.The face detection belongs to one kind of non-contact biological feature detection,which,as the first part of the face recognition,is such a process in which it searches in any image that is given through some strategy so as to determine whether there is a face or not.In such a society that pursues the efficiency,more and more real-time information is saved by way of image and video.Therefore,the face recognition technology is also widely applied to the personal identification and safety certification fields.Based on Adaboost algorithm,this article optimized algorithm learning and algorithm detection respectively and applied it to the face detection.The research is as follows:Firstly,the research status of the face detection technology at home and abroad was analyzed the common face detection methods were discussedSecondly,the Haar features,integral image and classifier as well as the principle and features of training face database were discussed deeply,and the reason of low detection rate in original Adaboost algorithm was also analyzed.As to the problems including the incorrect detection to inclined faces caused by the single training sample and excessive reliance to the training samples,the article puts forward two types of new feature templates and riches the training sample set in the aspects of mirroring.Thirdly,this article analyzed the reason of the sample weight imbalance caused by the existence of a large number of week classifiers in the training samples,and put forward the new calculation rule of the sample weight,which greatly increased the detection speed and detection performance of Adaboost algorithm.Lastly,the core theory features of PCA model was analyzed,a system of face detection and recognition was constructed using the optimized Adaboost algorithm.