Research And Implementation Of Facial Recognition System Based On WeChat Platform

In recent years, with the integration of computer and the rapid development and application of mobile Internet, smart mobile devices have began to come into our daily life, and gradually changed our habits of study and life. So-called smart mobile devices are refer to the smart phones, tablets, and some small mobile terminals, which have high performance hardware and independent operating system. And WeChat, as the popular application of mobile terminal, has developed into a "APP" platform, all kinds of native APP has been replaced with the WeChat, become one of WeChat components or reduced to the public accounts.Face recognition is a hot field of computer technology, its implementation in mobile terminal has been paid more and more attention.The author of this paper conducts on the study of face recognition technology and WeChat platform secondary development, then designs a face recognition system based on WeChat platform.The main research contents are summarized as follows:(1) Before face detection,some of color images with more complex characters background, using SWRW segmentation algorithm to pick up the characters from the background in order to avoid the interference of pseudo facial region.In image preprocessing phase, using two kinds of methods about image gray processing and histogram equalization, to remove or reduce the external environment disturbance to the image.(2) The research of face detection is using Ada Boost algorithm to select samples of Haar features, thus training classifier, and by adding Haar_T characteristics with up,down, left and right four directions to improve the accuracy of the classifier.(3) Facial recognition, including face identity identification and facial expression recognition, the main difference is feature extraction of both. Identity identification used the WSS-LBP operator to extract the local texture feature, received the highest recognition rate of 95% by conducting experiments on ORL face database; Expression recognition is on the basis of face detection, using the trained cascade classifier to select samples of Haar, Haar_T characteristic vector and the experiment on JAFFEfacial expression library, got 82.8% of the expression recognition rate.In the system implementation stage, the research needs to use OpenCV SDK toolkit for the development of functional modules, and compiles each algorithm into dynamic libraries, then by calling the compiled dynamic libraries to develop the WeChat public account. This way of partition between account development and algorithm development, speeds up the writing of program code.