Using Multimodal Biometric Fusion for Watermarking of Multiple Images

Recently, digital images are an essential source of data obtained from consumer devices, thus playing a crucial role in various important scenarios such as consumer apps, businesses, e-commerce, education, entertainment and healthcare. However, security and privacy are primarily concerned with the transmission and storage of these images. Therefore, there is an urgent demand to protect their copyright and prevent leakage. To address these concerns, this paper presents a novel multimodal biometric, encryption and watermarking-based method for digital image security. First, multimodal biometric features are extracted and fused using the customized deep learning model. Second, different secret keys are used to encrypt the fused features of the biometric images. Here, the fused features, considered as watermarks, are divided into sub-features before encrypting them with different keys. Third, to improve security while maintaining imperceptibility, the encoded sub-features are embedded into multiple cover media-based convolutional neural network (ConvNet). The experimental results demonstrate that the proposed system is highly secure and that it can recover complete marks under different attacks. Further, experiments illustrate the superior performance of our proposed system in terms of both imperceptibility and robustness compared with the competing schemes. It indicates a considerable improvement in imperceptibility and robustness of 62.83% and 25.16%, respectively over existing schemes.