رشا خضير رجه
Payment Systems play a vital role in the economy of a country. A payment system is used to settle financial transactions through the transfer of monetary value and consists of the various mechanisms that facilitate for the transfer of funds from one party to another.Payment systems are used to conduct financial transactions that are vulnerable to theft or attack. However, biometric payments, such as face recognition, may appear to be a viable solution. Face recognition is an approach of biometric techniques that is used to identify and recognize facial features in humans.. Face recognition consists of three phases: face detection, feature extraction, and the recognition phase (classification). Accuracy is not the only factor that determines the performance of face recognition. Time is also a crucial concern in real-time applications. Time is regarded as the most significant challenge in real-time environments. As a result, the proposed system focuses on proposing an improved face recognition payment algorithm by addressing the time issue associated with the performance of automatic face recognition processing. The proposed system suggests several algorithms to solve the time problem, as follows: The ViolaJones and skin color detection algorithms are used for the face detection phase. Frontal face images are processed using the Viola-Jones algorithm, while the right and left sides of the face are processed using skin color detection. For the feature extraction phase, the Discrete Cosine Transform (DCT) and Principal Component Analysis (PCA) are used. The multiple stages of reduction increase the speed of the entire system while decreasing complexity, processing time, and memory usage. A Support Vector Machine is used for classification (recognition phase), which has the ability to make a decision whether the person is known or not. Finally, we connect the system to an account for the payment system in which the user must enter the username and password to be certain that they are authenticated. If the user is authenticated, enter the amount of money to complete the transaction, or else cancel the payment process. The Olivetti Research Laboratory (ORL) and Head and Pose Images (HPI) datasets are used in the proposed system to evaluate the results. The proposed system reduces the execution time of the whole system and enhances the recognition rate as compared with other algorithms, such as using Viola-Jones with Linear Discriminant Analysis (LDA).The total time is reduced by half when using parallel processing. The total time without parallel processing is (1.226 sec) but in parallel processing, it II is (0.697 sec). The proposed system also enhanced the recognition rate (accuracy). The accuracy without parallel processing is 94%, but with parallel processing, it becomes 96%.
