International Journal of Science, Engineering and TechnologyAuthors: S.Hemanth, B.Shymala Devi
Abstract: In the ever-growing world of digitalization, securing and protecting the privacy of biometric information in cloud-based data sets have become central concerns. This work provides a secure representation of the blockchain technology and incorporation of sophisticated iris recognition technology to allow a method of a secure and non-repudiable biometric identification in the cloud condition. With the help of the MMU Iris Dataset, comprising 460 datasets of iris images of 46 individuals, the proposed system will deploy the utilisation of the pre-processing methods that may involve segmentation, normalisation, and denoising to improve the quality of the images. Discriminative and compact feature vectors are produced by applying deep learning techniques such as InceptionV3 and classical Log-Gabor filters on feature extraction. To maintain confidentiality and integrity, these vectors are encrypted with the AES-256 and hashed with the SHA-256. The hash-coded templates are subsequently enrolled and published on a permissioned block chain network with smart contracts to receive identity verification, admittance restrictions and withdrawal. With homomorphic encryption, biometric templates may be securely matched, without decryption and without breaching an end-to-end privacy. The proposed model reached the recognition accuracy of iris definition of 97.8%, which was much better than the recognized models of recognition such as ResNet50 (95.7%) and VGG16 (94.1%). The system also has a low encryption latency and has a high blockchain throughput (1200 TPS) which is scalable to the real world. This framework exhibits a combination of the benefits of biometric authentication and decentralization of blockchain, thus meeting both of the criteria of cloud-based identity management data security and trustworthiness. The proposed model can, therefore, offer a trusted, transparent and privacy-wise solution to next-gen biometric systems.
