Mobile payment recognition technology based on face detection algorithm

摘要

In view of theproblem that thetraditional face recognition algorithm cannotbeeffectively appliedto the identification of mobile payment, a lightweight improvement scheme of the dynamic heteroscedasticitybased on the classical scale transformation algorithm is proposed, which canautomatically study and adaptively add the reliable test samples to the training sample space,with reasonable division and scientific weight distribution characteristics, so that the scheme canimprove both the recognition rate and running time. The improved algorithm is tested in theORLface database and Yale face database, respectively, with the recognition rate improved by 6.13%and 14.11%, and the operating efficiency increased by 9.1% and 4.7%, respectively, comparedwith the classic scale transformation algorithm. The recognition rate in the ORL face databasehas achieved 74.05%, with significant improvement compared with PCA, LBP, and other classicalgorithms. And the mobile payment scheme is implemented inAndroid smart phones. The experimentaldata has verified the feasibility of the improved algorithm in the Android system. Finally,an improved scheme based on the cloud architecture is proposed.