Aiming at the problem that traditional methods can only increase the resolution of local regions in face images

and the whole face image passivation blur is also present

a face super-resolution method based on generative adversarial networks is proposed. Through the PSNR value

SSIM value and MOS test comprehensive considerations

it is concluded that the face super-resolution technology based on this method has great advantages compared with other traditional image processing methods or deep learning methods such as SRCNN

DRCN

and ESPCN

etc

the super-resolution face generated by this method not only improves the magnification

but also obtains better high-frequency details of face.