.png){kind=icon}
.png){kind=icon}
Return
Improving iris biometric pattern super-resolution using edge focused generative adversarial network
.png){kind=icon}
PRE
AI
DOI：10.1016/j.bspc.2026.110520.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
There are different conditions for capturing iris biometric images in various authentication systems. In case of encountering low-quality cameras, environment light problems or other high-quality image capturing problems, image super-resolution can help for having a better authentication process with low-resolution iris images. Different methods for Image SR presented for different purposes in recent research and SRGAN is one of the most operational ones. In this work a new method was proposed for use simultaneously with the SRGAN to make the edges of iris pattern images much more detailed. By using this method low-quality iris patterns be super-resolved in a way that can be used as iris code for the authentication process. The accuracy of the system increases and FAR and FRR parameters decrease in a way that the overall system is much more reliable. SR iris pattern image accuracy increased noticeably in comparison to the LR iris pattern image. In addition, we produced natural LR images for this approach using the deep neural network(DNN) to have a natural iris pattern dataset for learning the GAN deep neural network. This process results in more natural quality enhancement for the iris pattern SR procedure.
Keywords:
iris biometrics
super-resolution
generative adversarial network
edge enhancement
authentication system
Journal
IF:
4.9
Papers:
9.8K
Citations:
2.4W
Organization
No organization information available