Rotation Invariant Fingerprint Detail Point Feature Extraction Network Based on Attention Mechanism
Accurately extracting the details of fingerprint images is a key step to achieve accurate fingerprint matching.Detail point features often appear in any direction,and ordinary convolution neural networks do not explicitly model rotational transformations.In order to solve the above problems,this paper proposes a rotational invariant fingerprint detail point feature extraction network based on attention mechanism.In this model,the weight sharing of rotation is added to the model design,so as to reduce the number of model parameters,fundamentally enhance the generalization ability of the model,and improve the computational efficiency.This study provides a new idea for constructing a stable and efficient fingerprint detail point detection system.
万方数据
维普
