基于NSCT和PCNN的红外与可见光图像融合方法

打开文本图片集

中图分类号：V445 文献标志码：A 文章编号：2095-2945（2025）27-0042-04

Abstract：Aimingattheproblemsofblurredboundariesandlosttextureduringmultimodalfusionofinfraredandvisible images，afusionmethod basedonnon-subsampledcontourlettransform（NSCT）andpulse-coupledneural network（PCNN）is proposedtoimprovethefusionquality.First，twosetsof sourceimagesare decomposedintohighandlowfrequencysubbands byusingNSCT，andthenfusionrulesbasedonPCNN-areintroducedtofusethehighandlow frequencysubbands.Fialy， inversionreconstructionisapliedtoobtainthefinalfusedimage.Experimentalresultsandobjectiveevaluationshowthatthe fusionmethodinthispaperissuperiortoothercomparisonmethodsandcanefectivelyretaininfraredimagefeaturesandvisible lightimagetextureinfomation.Experimentalresultsshowthatthismethodissuperiortotecomparisonalgorithmintermsof subjectivevisualqualityandbjectiveevaluationindicators（EN=7O1，VIF=0.719），andcanfectivelybalancethecotradictio between maintaining the saliency of infrared targets and extracting visible light texture details.

Keywords：image fusion;non-subsampledcontourlettransform;pulse-coupledneural network;infraredimage;visibleimage

红外和可见光图像融合作为一种多模态融合技术，近年来广泛应用于建筑、电力检测和医学成像等领域。（剩余4564字）