Quality Assessment for Fingerprints Collected by Smartphone Cameras

We propose an approach to assess the quality of fingerprint samples captured by smartphone cameras under real-life scenarios. Our approach extracts a set of quality features for image blocks. Without needing segmentation, the approach determines a sample's quality by checking all image blocks divided from the sample and for each block a trained support vector machine gives a binary indication - "high-quality" or "non-high-quality" (including the low quality case and the background block case). A quality score is then generated for the whole sample. Experiments show this approach performs well in identifying the high quality blocks - the Spearman correlation coefficient between the proposed quality scores and samples' normalized comparison scores (ground truth) reaches 0.53 while the rate of false detection (background blocks judged as high-quality ones) is still low as 4.63 percent over a challenging dataset collected under various real-life scenarios.