Automated Fast Anti-Spoof Biometry Using Random Temporal History and Numerical Indexing Based Biospeckle Analysis

Abstract

In recent years, we have witnessed the adoption and evolution of biometric technology in various versatile applications such as access control, surveillance, and forensic investigations. However, one of the constraints in conventional biometric systems is the high possibility of the system being tricked or spoofed by forged biometrics, such as by artificial samples, photographs, video clips, etc. To identify such a spoof attack for fingerprint biometry, in this paper we propose an automated biospeckle based fast, simple, low-cost, and full-field technique. The specimen is illuminated by collimated light from a coherent laser source. The back-reflected light from the specimen is captured by a CCD camera, resulting in the formation of a dynamic granular shaped speckle pattern on the detector screen. To avoid the false results generated due to noise and vibrations, a spatial support is used. For automated processing of the recorded time-series images to estimate the dynamicity, an efficient and fast numerical biospeckle analysis strategy is introduced by combining statistical temporal history, numerical indexing, and region of interest selection strategy. Simulation studies confirm accuracy, high speed, and memory compactness of the proposed technique. It has been established that the proposed strategy is proficient in testing genuineness of the biometric traits. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.