Alignment-Free Cancelable Biometric: A Contemporary Survey, Opportunities & Challenges
Pratima Sharma, Gurjit Singh Walia, Rajesh Rohilla · 2020
This survey aims to review different alignment free recognition approaches inclusive of advantages and limitations of various state-of-the-art biometric template protection schemes. Occurrence of rotational and translational variation while acquiring images followed by one way transformation or cryptographic hashing, results in degradation of biometric recognition performance. The non-invertible transformation and cryptographic methods addresses the security and privacy threats associated with biometric templates. Although, maintaining high recognition accuracy became the foremost challenge for various proposed approaches of biometric template protection. Among largely suggested approaches for performance enhancement, need for registration or alignment of biometric features came out as a difficult requirement to be fulfilled. Hence, performance degradation mainly due to elastic distortion can be confronted using various alignment-free approaches. The alignment-free method involves use of a feature extraction process that extracts invariant attributes found in local geometrical structures of biometric templates. This survey article discusses various techniques of forming local structure accompanying non-invertible methods to generate alignment free cancelable templates or bio-cryptosystems. Further, a comparative analysis of various alignment-free cancelable approaches are compiled in tabular manner. Review also includes a summarization of the research gaps to aid in finding future directions.