A Study Of A Biometric Identification Technique Based On Naked Footprint

The theft of an individual’s biometric feature especially in fingerprint system where animpostor creates a prosthetic fingerprint or biometric characteristic to launch an attack on thesystem is one but many drawbacks of notable biometric systems currently in application.Footprint identification which can be defined as the measurement of footprint features forrecognizing the identity of a user has surfaced recently. The shoe wearing culture of mostpeople has made it arduous for impostors to obtain footprints for forgery attacks; footbiometric is therefore the new biometric trait that is more reliable and less vulnerable againstspoof attacks. Footprint is projected to be the answer to the continuing threats to critical or highsecurity infrastructure and the growing national security vulnerabilities.This paper presents a biometric approach for personal identification using static footprintfeatures viz. friction ridge/texture and foot shape/silhouette. The method works such that,naked footprints of users are captured from volunteers.The scanned images undergo preprocessing to convert them from their initial RGB stateto gray scale and subsequently are binarized. The shape feature is extracted via the GradientVector Flow (GVF) snake model. The corresponding minutiae extraction is executed usingthe Crossing Number (CN) method. The minutiae extraction operation deduces the minutiaelocation and the minutiae angles. This stage marks the end of the feature extraction processesrespectivelyMatching is then effected in a parallel manner based on these two features. The shapematching is preceded by a further feature extraction where the shape extracted using the GVFsnake algorithm is subsequently converted to a series of codes via chain code algorithm.Matching the two series derived from the foot shapes is achieved using a method calleddynamic time warping (DTW). The sequences derived from the foot shapes are warped in anonlinear manner and the measurement of the DTW distance is taken based on an optimalwarping trajectory of two sequences to execute a match.The texture or minutiae matching is based on the Miniature Score Matching (MSM).During the matching process each input minutiae point is matched against a correspondingtemplate minutiae points. In each case, the template and input minutiae are selected asreference points for their respective data sets.The scores emanating from the two features are normalized via min-max normalization totransform them into a common domain. Subsequently the individual matching scores arecombined using the sum rule fusion to produce a single scalar score which is then used toderive the final decision. The experimental results show a high degree of accuracy withverification clocking98.97%accuracy and imposter verification of25%. The algorithms used provided convincing matching results to establish that the nakedfootprint is a credible biometric feature.