SlapSegII Analysis: Matching Segmented Fingerprint Images

摘要

This report is an extension of the SlapSegII evaluation which measured the ability of slap segmentation algorithms to successfully segment four-finger slaps into individual fingerprint images. This study compared the matching accuracy of hand-marked segmentation and automated segmentation algorithm results to examine the effect of automated slap fingerprint segmentation errors on matching accuracy. A couple other issues this study examined are: 1) How detrimental was excluding fingerprint ridge structure (over-segmentation) to matching. 2) How much did fusion of multiple fingers compensate for segmentation errors. The only dataset from SlapSegII with mates available for matching was the dataset with 2-inch high slaps, so the study looked at matching accuracy for single finger, two index fingers and eight fingers (no thumbs).