Face recognition encryption authentication method and system combining watermark and deep learning

Abstract

The invention relates to a face recognition encryption authentication method and system combining watermarking and deep learning, and belongs to the field of face recognition authentication. A camera collects a face image and embeds a watermark to obtain a watermarked face image; the client extracts the watermark information and compares the watermark information with the watermark image for authentication; the client side adopts a deep learning algorithm based on FaceNet to extract face features of the successfully authenticated face image, and adopts a CKKS fully homomorphic encryption algorithm and a national cipher SM4 algorithm to encrypt the face features to obtain face feature ciphertext data; the database server calculates a Hamming distance between the encrypted face feature template and a pre-stored face feature template, and performs encryption to obtain Hamming distance ciphertext data; and the identity authentication server decrypts the Hamming distance ciphertext data and compares the Hamming distance plaintext value with a preset Hamming distance threshold value to obtain a face recognition authentication result, so that the security of a face image data source can be ensured, and the recognition accuracy and efficiency are improved.

Description

technical field [0001] The invention relates to the field of face recognition authentication, in particular to a face recognition encryption authentication method and system combining watermarking and deep learning. Background technique [0002] The main idea of ​​the current face recognition encryption authentication method is to encrypt the extracted face features, complete the face recognition authentication in the ciphertext domain, and the authentication is successful if the recognition is successful. However, the following problems generally exist in the existing commonly used face recognition: (1) problems in security. Malicious attackers can bypass the camera and directly provide problematic face images to the feature extraction module, or tamper with the face images during the process of sending them to the feature extraction module, which cannot guarantee the reliability and security of the source of face image data ; (2) Performance issues. The entire process of...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to provide a face recognition encryption authentication method and system combining watermarking and deep learning, which can ensure the security of the source of face image data, and improve the recognition accuracy, and solve the problem that the source of face image data cannot be guaranteed in the prior art security, problems with low identification accuracy

Images