Privacy preserving steganography based biometric authentication system for cloud computing environment

Recently, cloud computing (CC) received significant interest among organizations and individuals. Despite the significant benefits of CC, security and privacy are considered as the major issue. Biometric authentication is commonly employed for authentication purposes and has gained attention among several researchers due to its stable and high recognition rate. Amongst the several biometric authentication models, fingerprint is treated as an effective one to achieve security and privacy. Besides, image steganographic approaches are applied in order to enhance the security of the biometric data. The state of art biometric data hiding techniques generally performs the data embedding on the region which does not encompass key features of the biometrics. With this motivation, this paper presents a privacy preserving steganography based biometric authentication system (PPS-BAS) for cloud environments. The goal of the PPS-BAS model is to hide the fingerprint image (secret image) into the eye retina image (cover image) and transmits it to the cloud in an encrypted way. The proposed PPS-BAS model involves multilevel discrete wavelet transform (DWT) technique to split the cover image in order to identify the pixel location. Besides, continuous pigeon inspired optimizer (CPIO) algorithm is applied to determine the optimal pixel points in the cover image. At the same time, a Q-learning technique is employed to extract the minutiae from the fingerprint image and is then hidden into the optimal pixel locations in the cover image. In order to further increase security, double-logistic chaotic map (DLCM) model is applied for encrypting the stego image which is then transmitted to the cloud server. After the reconstruction of the original fingerprint image (secret image), the biometric recognition process takes place using the Scaled Conjugate Gradient (SCG) based back propagation neural network (BPNN) model. A detailed simulation analysis is performed to highlight the enhanced outcomes of the proposed PPS-BAS model and the comparative results analysis ensured the betterment of the PPS-BAS model over the recent state of art biometric authentication systems. © 2022 The Authors