Pollen　classification　is　crucial　for　the　prevention　of　pollen　allergy　diseases.　Light　Microscope　(LM)　based　pollen　classification　methods　have　received　more　attention.　However,　a　large　number　of　LM　images　in　the　real　world　have　blurring　problems,　which　seriously　affect　the　classifier＇s　classification　accuracy.　To　address　these　problems,　an　image　quality　assessment　(IQA)　method　is　considered　to　pick　high-definition　pollen　datasets　for　aiding　in　solving　classifier＇s　low　accuracy　problem.　Unlike　existing　image　quality　assessment　methods　that　are　commonly　based　on　global　features,　we　focus　on　the　local　texture　information　of　pollen　grains.　We　propose　an　image　quality　assessment　method　for　small　pollen　grains　in　pollen　images,　which　consists　of　three　main　modules.　First,　in　the　global　IQA　model,　Siamese　network　is　applied　to　focus　on　the　coarse-grained　image　quality　information.　Secondly,　in　the　Local　IQA　module,　pollen　grain　fine-grained　texture　information　is　considered　and　incorporated　into　the　image　quality　information.　Finally,　in　the　IQA　fusion　model,　Global　IQA　and　Local　IQA　are　integrated　based　on　a　weighting　strategy.　On　the　IQA　pollen　dataset,　this　method　showed　a　13.8%　improvement　in　accuracy　compared　to　well-expressed　quality　assessment　methods.　To　verify　the　effectiveness　of　this　paper＇s　quality　assessment　method　applied　to　the　classification　problem,　we　compare　the　classification　performance　of　introducing　different　quality　assessment　methods.　The　classification　method　with　the　introduction　of　this　paper＇s　image　quality　assessment　improves　the　average　accuracy　by　3%.　©　2023　IEEE.