Facial and Palm-Based Biometric Authentication and Data Security

Abstract Biometric Authentication (BA) is frequently used for authentication owing to its high recognition rate. The existing biometric hiding algorithms execute data embedding on areas that do not encompass key features of the biometric. Moreover, these techniques lacked authorization. Thus, a Secure Data Transfer model with BA and Blockchain (BC)-based authorization is proposed. Primarily, the data owner registers their details and the registered face and palm image undergoes pre-processing. By employing Pruned Residual Network 50 (PRESNET 50), the facial landmarks are extracted from the pre-processed face image. Next, Digit Folding based Log Facial Jaw Points Curve Cryptographic (DF-LFJPCC) is executed based on the jaw points to generate a secret key. Then, the Tan Sigmoid-based Convolutional Neural Network (TS-CNN) classifier is trained with the features of the pre-processed images and facial landmarks. After registration, the user logins, and their processed face and palm features are given to the trained TS-CNN for authenticating the user. The secret is also used to improve the authentication process. After successful login, the file to be uploaded is converted into cipher, which is then encrypted using Log Facial Jaw Points Curve Cryptographic (LFJPCC) and uploaded to the cloud server. In the end, authorization is performed in the BC based on the hashcode generated using Faro shuffle -Tiger (FS-Tiger) when a user requests data. As per the experimental analysis, the proposed technique outperforms prevailing models.