Current quantitative approaches to fingerprint matching and analysis are not based on human data and therefore do not take advantage of the full capabilities of the human visual system. Since humans routinely outperform automated fingerprint recognition systems, it is clear that quantitative approaches can be improved by adopting some of the strategies that humans employ; however, humans often have difficulty describing the result of perceptual processing, and may not even know what information they are using. To address this deficit, the authors used eye tracking to identify what information human experts rely on. They constructed a portable eye tracking system that enabled them to collect data from experts and novices while they perform tasks similar to latent print examinations. Once they analyzed the data, they obtained a record of the regions visited by the experts as they compared pairs of fingerprints. The authors then developed a series of computational analyses to identify the nature of the expertise. This took the form of data reduction procedures on pixel crops from the fingerprint images, as well as the development of candidate information metrics that the data from experts helps validate. The results demonstrate clearly that human expertise can be inferred from eye gaze information through a process of carefully designed studies and hypothesis testing of candidate information metrics. Because the authors' candidate metrics take the form of mathematical and computational models, they are readily applicable to machine comparison approaches, and also can be used to identify the diagnosticity and rarity of particular features in novel prints. Appendix, figures, and references (Author Abstract)
Downloads
Similar Publications
- Garden Grove Police Department Public Safety Dispatcher Training and Evaluation Guide
- Syndrome-informed Phenotyping Identifies a Polygenic Background for Achondroplasia-like Facial Variation in the General Population
- Improving the Effectiveness and Utilization of Neighborhood Watch Programs: Executive Summary