Image　deblurring　is　a　classical　problem　in　the　field　of　computer　vision.　The　purpose　of　this　paper　is　to　restore　the　multi-layer　pollen　image　into　an　image　with　clear　texture　when　the　fuzzy　kernel　is　unknown.　In　this　paper,　a　pollen　texture　migration　deblurring　network　based　on　self-coding　generator　is　designed,　which　uses　the　maximum　similar　texture　information　between　adjacent　layers　of　pollen　to　recover　clear　pollen　layer　by　layer.　A　pollen　image　deblurring　method　based　on　texture　migration　DeblurGAN　of　countermeasure　generation　network　is　proposed.　The　texture　migration　network　is　used　as　the　recovery　network,　and　the　residual　pair　information　is　extracted　by　the　residual　block　in　the　encoder　to　simplify　the　training.　Secondly,　the　jump　connection　layer　is　used　to　retain　the　information　to　remove　the　color　offset,　and　the　deconvolution　layer　of　the　generator　is　improved　into　the　alternating　layer　of　nearest　neighbor　interpolation　and　convolution,　which　can　help　to　remove　the　checkerboard　error.　Finally,　a　layer　of　convolution　is　used　to　reconstruct　the　feature　image.　The　algorithm　is　evaluated　on　multi-layer　pollen　data　sets.　From　the　evaluation　index　and　subjective　effect,　it　can　be　seen　that　the　method　proposed　in　this　paper　has　stronger　image　restoration　ability,　retains　rich　texture　information,　and　can　effectively　improve　the　image　deblurring　effect.　©　2022　IEEE.