基于改进条件扩散模型的残简文字修复方法

打开文本图片集

关键词：秦简文字；图像修复；扩散模型；空间通道重建卷积；噪声估计网络；注意力机制中图分类号：TN911.73-34；TP391.41 文献标识码：A 文章编号：1004-373X（2025）17-0061-07

Abstract：Qinbambo slipcharacter wereburiedundergroundforalong time，andtheir surfaces were erodedbyrainwater， soalargenumberof brokenbambooslipsappearedduring theprocessof excavation.Inviewof theunclearcharactersand missing fontstructureinformationonbambooslips，amethodforrepairing brokenbambooslipcharacters basedonimproved conditionaldifusionmodelisproposed.Byoptimizingthenoiseestimationnetwork，theglobalstructureinformationandlocal texturedetailsofQinbambooslipcharactersarecaptured，soastoobtainbeterimageinpaintingefect.Intheoptimized network，aspatialchannelreconstructionresidualmodulewithdualinformationinputisintroducedtoimprovethetextureclarity oftherepairedimagefirst，andthentheglobalcontextnetworkGCNetisintroducedtopropagatelocaltexturefeaturestoa globalrange，soastostrengthenthedependencybetweenpixelsofcharactersfarapartandmaketherepairedtextstructure complete.Thecomparisonexperimentswerecariedoutbetweentheproposed modelandtheotherimagerepairmodelsonthe self-madeQin bambooslipcharacterdataset.Thesimulationresultsshowthattheproposedmethodachievesgoodresultsin subjectivevision，withpeaksignal-to-noiseratio（PSNR）andstructural similarity（SSIM）reaching34.12dBand97.8%， respectively，whichare1.68dBandO.6%higherthanthoseofthedifusionmodelPalete，respectively.Tosumup，the inpainting effect can meet the visual needs of archaeologists，philology experts and historians.

Keywords：Qin bamboo slipcharacter;image inpainting;difusionmodel;spatial channelreconstruction convolution;noise estimation network;attentionmechanism

0 引言

秦简文字记载了秦朝的法律制度、行政文书、医学著作等，为中国书法、秦朝政治、法律、文化、医学等方面的发展历史提供了详实的资料，具有重要的历史意义和学术价值。（剩余9144字）