Sparse Representation-Based Classification for Enhanced Biometric System Reliability

In an increasingly digitized world, cybersecurity has emerged as a critical concern driven by technological advancements and the extensive digitization of various activities. The protection of computer systems, networks, and data from cyber threats is essential to mitigate adverse impacts on individuals, businesses, and governmental entities. Biometrics represents a crucial component of cybersecurity, offering solutions to prevent unauthorized access, identity theft, and illicit data modifications. This paper introduces an innovative contactless biometric recognition system that employs Histogram of Oriented Gradients (HOG) and Binarized Statistical Image Features (BSIF) for feature extraction, and an enhanced version of Extreme Learning Machine (ELM) termed Rough ELM (RELM), inspired by Sparse Representation Classification (SRC). Experimental results using standard palm print databases demonstrate a significant improvement in identification rates. The proposed system integrates RELM methodology to enhance accuracy and efficiency by combining traditional handcrafted techniques with extreme learning features. The findings highlight the effectiveness of this approach.