Quantiative Conformance Testing for ICAO Standard on Facial Photo Specification

초록

With the aim of verifying if facial photos meet ICAO requirements; this paper proposes effective metrics for quantitative conformance testing for ICAO standards on facial photo specifications. In order to conveniently utilize in facial photo quality validation systems, each metric is normalized to [0, 10] as a quality score of each requirement. Furthermore, to automatically assess the quality of facial photos, this paper proposes a robust method of extracting necessary features in the images for the conformance testing. The proposed extraction method takes the advantages of combining color, intensity and edge information. Experiments were performed on a subset of the FERET database, CVLab database and GTAV database. The experimental results demonstrated the accuracy and robustness of the proposed metrics and extraction method.