Personal Identification from Lip-Print Features using a Statistical Model
This work addresses biometric identification for security applications, but it is incremental as it applies existing techniques to lip prints.
The paper tackles personal identification by analyzing lip-print groove features using edge detection and connected-component analysis, achieving satisfactory results with metrics like FAR, FRR, and ROC for biometric verification systems.
This paper presents a novel approach towards identification of human beings from the statistical analysis of their lip prints. Lip features are extracted by studying the spatial orientations of the grooves present in lip prints of individuals using standard edge detection techniques. Horizontal, vertical and diagonal groove features are analysed using connected-component analysis to generate the region-specific edge datasets. Comparison between test and reference sample datasets against a threshold value to define a match yield satisfactory results. FAR, FRR and ROC metrics have been used to gauge the performance of the algorithm for real-world deployment in unimodal and multimodal biometric verification systems.