Image Rescaling with the Ruler on Latent Fingerprint Images

In this study, a novel method towards rescaling latent fingerprints with the help of ruler is presented. The proposed method is developed using traditional machine learning algorithms and deep learning. This method includes segmenting the ruler region, finding the distance between the ruler lines, and determining the scale. The ruler segmentation model used in the proposed method has 99.65% precision and 91.03% sensitivity. In addition, the use of fingerprint images whose scale is changed by the proposed method in matching algorithms increases the matching performance by 12.5%.