Fingerprint sensors often have difficulty authenticating users if the user’s fingers are wet or moist, e.g., due to sweat. This disclosure presents robust techniques for fingerprint identification based on the K-SVD algorithm, which is a technique to represent images in a sparse manner. A K-SVD dictionary is created out of enrolled fingerprint images. A fingerprint that is to be authenticated is segmented into blocks, and each block is projected against the dictionary. A heat map of highest projection coefficients is formed, and overall match-score is calculated. The overall match-score is used to authenticate the fingerprint. The dictionary stores the essential features of the enrolled fingerprints, and the enrolled fingerprint images are deleted. Fingerprint authentication is made possible without actual storage of the enrolled fingerprints, which serves to improve security.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Sammoura, Firas and Robbins, Matthew, "Authentication Using Sparse Modeling of Fingerprint Images", Technical Disclosure Commons, (October 08, 2018)