Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This study evaluates the effectiveness of Open Face Image Quality (OFIQ) measures for assessing the quality of ID card images used in remote verification systems. By implementing a preprocessing pipeline that includes corner detection, perspective normalization, and foreground masking, the authors ensure accurate quality measure computation. The findings reveal that certain OFIQ measures can significantly enhance the performance of presentation attack detection (PAD) algorithms across various ID card datasets, indicating a potential improvement in security for remote verification processes.
Quality assessment using OFIQ measures can notably boost the effectiveness of presentation attack detection in ID card verification systems.
This paper addresses the challenge of assessing image quality in ID cards in remote verification systems by applying capture-related quality measures from the Open Face Image Quality (OFIQ) standard to ID card images. Our preprocessing pipeline includes corner detection, perspective normalization, and comprehensive foreground masking to ensure accurate and unbiased quality measure computation. We evaluate the effectiveness of these measures by analyzing their correlation with the performance of three presentation attack detection (PAD) algorithms across four diverse ID card datasets, where two datasets contain bona fide, i.e. pristine, images and two contain printed mock ID cards. Our results suggest that quality assessment based on some OFIQ measures can significantly improve PAD performance.