超级缩放解析内核Blind SuperResolution Kernel Estimation using an InternalGAN
Abstract
Superresolution(SR)methodstypicallyassumethatthelow-resolution(LR)image
wasdownscaledfromtheunknownhigh-resolution(HR)imagebyafixed‘ideal’
downscalingkernel(e.g.Bicubicdownscaling).However,thisisrarelythecase
inrealLRimages,incontrasttosynthetically
暂无评论