Fingerprint Reference-Point Detection

摘要: A robust fingerprint recognition algorithm should tolerate the rotation and translation of image. One popular solution is to consistently detect a unique reference point compute orientation for translational rotational alignment. This paper develops an effective locate corresponding accurately all types fingerprints. To reliable field, improved smoothing method proposed based on adaptive neighborhood. It shows better performance in filtering noise while maintaining localization than conventional averaging method. The reference-point multiscale analysis consistency search local minimum. computed differences between radial directions fromthe point, which are radii emitted from with equivalent angle interval, ridge orientations along these radii. Experimental results demonstrate that our can high accuracy