Deep convolutional neural network for latent fingerprint enhancement

作者：

Highlights：

•

• A better strategy for latent fingerprint data augmentation is proposed to train CNN.

• FingerNet is proposed with the pixels-to-pixels and end-to-end learning manner.

• Multi-task learning and residual learning strategies are studied thoroughly.

• Competitive matching performance is achieved with faster inference speed.

摘要

•A better strategy for latent fingerprint data augmentation is proposed to train CNN.•FingerNet is proposed with the pixels-to-pixels and end-to-end learning manner.•Multi-task learning and residual learning strategies are studied thoroughly.•Competitive matching performance is achieved with faster inference speed.

论文关键词：Latent fingerprint enhancement,Convolutional neural network,Pixels-to-pixels and end-to-end learning,Multi-task learning

论文评审过程：Received 26 January 2017, Revised 25 June 2017, Accepted 18 August 2017, Available online 24 August 2017, Version of Record 28 September 2017.

论文官网地址：https://doi.org/10.1016/j.image.2017.08.010