We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Verifiable speech retrieval algorithm based on diversity security template and biohashing.
- Authors
Zhang, Yuan; Huang, Yi-bo; Chen, De-huai; Zhang, Qiu-yu
- Abstract
In order to solve the problem of leakage in biometrics, to improve the performance of speech retrieval and the security of single biometric template, this paper proposes a verifiable speech retrieval algorithm based on diversity security template and BioHashing. Firstly, both of biometrics which are the cross-correlation cosine and improved sub-band energy entropy ratio are fused to construct the time-frequency biometric vectors, finally the vectors are classified by the K-medoids algorithm. Then, the dimension vectors of the orthogonal set matrix which is constructed by Schmidt orthogonalization of the QCNN(Quantum Cellular Neural Network) chaotic matrix are one-to-one corresponding to the biometric vectors of the corresponding class to form the diversity biosafety templates, and the templates are further quantified to generate BioHashing sequences. Finally, the address index of dimension vectors of QCNN chaotic matrix and BioHashing sequences of corresponding class are one to one correspondingly encrypted by QCNN chaotic mapping encryption to construct hash index, and then the hash index is stored in the system hash index table of the cloud server. At the same time, the encrypted speech is uploaded to the cloud sever after 2D-LICM chaotic scrambling-shift encryption which is convenient to match and retrieve data for the users. Experimental results show that the diversity template has better security and complexity by using K-medoids algorithm to classify the biometric. Hash index and encrypted speech can effectively prevent the leakage of privacy data. At the same time, the retrieval system has better discrimination and retrieval performance, and it performs verifiable retrieval of speech for content preservation operations.
- Publication
Multimedia Tools & Applications, 2023, Vol 82, Issue 24, p36973
- ISSN
1380-7501
- Publication type
Article
- DOI
10.1007/s11042-023-14873-5