基于双混沌映射与神经网络的加密人脸识别鲁棒性算法研究

A Robust Algorithm of Encrypted Face Recognition Based on Double Chaos Map and Neural Network

针对传统人脸识别的安全性问题提出了一种基于双混沌映射与神经网络的加密人脸识别鲁棒算法.首先利用Logistic与Henon混沌映射产生密钥,其次利用密钥在DWT-DCT变换域内加密人脸图像,结合PCA与神经网络实现对加密人脸的识别,最后进行常规攻击、几何攻击和遮挡攻击来测试其鲁棒性.实验结果表明,该算法具有良好的加密效果与鲁棒性.

In the report,aimed at the security problems of the traditional face recognition method,a robust algorithm of encrypted face recognition based on double chaos map and neural network was proposed. Firstly,the Logistic Map and Henon Map were used to generate the key; Secondly,the key was used to encrypt face image in DWT-DCT transform domain,combined with PCA and neural network to recognize the encrypted face; Finally,the robustness was tested by conventional attack and geometric attack and occlusion attack. The results showed that the algorithm has good encryption effects and has strong robustness.