BriefGPT.xyz
Dec, 2018
基于M-Net卷积神经网络的指纹图像去噪和修复
FPD-M-net: Fingerprint Image Denoising and Inpainting Using M-Net Based Convolutional Neural Networks
HTML
PDF
Sukesh Adiga V, Jayanthi Sivaswamy
TL;DR
使用端到端可训练的卷积神经网络FPD-M-net，将指纹去噪和修复问题作为分割任务，以结构相似度损失函数为基础，从噪声背景中更好地提取指纹，并在Chalearn LAP修复竞赛的第三轨道中排名第3。
Abstract
The
fingerprint
is a common
biometric
used for authentication and verification of an individual. These images are degraded when fingers are wet, dirty, dry or wounded and due to the failure of the sensors, etc. T
→