作者: En Zhu , Chun-Feng Hu , Jian-Ping Yin , Guo-Min Zhang
DOI:
关键词:
摘要: Ridge orientation is one of the fundamental features a fingerprint image. Orientation estimation serves for feature extraction and matching base recognition. Most existing methods are based on characteristic pixel intensity in block. This paper uses neural network to learn ridge orientation. The trained has property responding true with large value false small value. When estimating orientation, responded values each at image block used compute field. proposed method turns out more robust than method.