A novel strategy for image segmentation of latent fingerprints

Digital image segmentation is one of the most important stages in the implementation of an Automatic Fingerprint Identification System. This work describes a strategy for image segmentation of latent fingerprints using a proper combination of operators achieving better results than other approaches reported in the literature. Latent fingerprint images are low quality images that make more difficult the segmentation process. The proposed segmentation strategy is based on the gradient magnitude of the image and the detection of regions. This strategy was implemented in Matlab and Java, and was tested using fingerprint images of the Fingerprint Verification Competition databases, which are commonly used for these purposes. The results achieved show a significant improvement compared with representative algorithms of literature, such as those based on the variance of image.