Privacy Preserving Face Identification in the Cloud through Sparse Representation

Jin, Xin; Liu, Yan; Zhao, Geng; Chen, Yingya; Guo, Kui

doi:10.1007/978-3-319-25417-3_20

Cited by 15 publications

(1 citation statement)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The larger the obtained score is, the higher the relationship between models. The system accessibility is guaranteed if the test person score exceeds a certain level (He et al 2013;Xiaodong et al 2019;Osadchy et al 2010;Jin et al 2015). In real-world biometric systems for a number of reasons, unauthorized models may generate higher scores than those of some authorized models.…”

Section: Verification Quality Metricsmentioning

confidence: 99%

Efficient chaotic-Baker-map-based cancelable face recognition

Faragallah

Naeem

El‐Shafai

et al. 2021

J Ambient Intell Human Comput

View full text Add to dashboard Cite

Biometric recognition schemes are commonly utilized for security purposes. These schemes face two major challenges: security challenge and reliance on a particular biometric for authentication. The protection challenge comes from the use of basic biometrics in records. Therefore, if these records are compromised, the biometrics will no longer be valid. Consequently, it is necessary to preserve basic biometrics by protecting them from being used in biometric records. Cancelable biometric recognition systems rely on changing the information or biometric features to different formats, so that persons can use their specific biometric models in single or multiple systems. The mixture of chaos theory and cryptography shapes a crucial area for information security. The newest development in encryption systems is chaos-based for various distinctive attributes such as sensitivity to preliminary conditions, non-convergence, non-periodicity, and control parameters. This paper introduces a method to produce numerous encrypted biometric templates that are re-created by various convolution kernels generated from employing chaotic Baker map in different domains. Our proposed method has superior performance in treating variations in illumination, occlusion and facial expressions. It also has the added novelty of being able to perform authentication in the encrypted domain. In addition, the same approach can be applied on different databases. The chaotic map effect in different domains is evaluated on the widely-used AT&T, YALE, UFI, LFW, and FERET databases. The cancelable biometric system using the proposed Discrete Wavelet Transform (DWT) domain encryption with various keys has the best performance among all other implementations. In average, we achieved a 2% error probability, a 0.3 s authentication time, a 0.02% False Rejection Rate (FRR), a 0% False Acceptance Rate (FAR), a 0.0% Equal Error Rate (EER) and a 98.43% accuracy. Our proposed method also provides template diversity.

show abstract

Section: Verification Quality Metricsmentioning

confidence: 99%