基于生成對抗網(wǎng)絡(luò)的服裝草圖轉(zhuǎn)換研究

打開文本圖片集

中圖分類號：TP391.4；TP18 文獻標識碼：A 文章編號：2096-4706（2025）08-0061-05

Abstract： The image conversion of clothing sketches has enormous application value in fields such as visual editing， clothingdesign，ndhuman-computerinteraction.Atpresent，mostofthejectsconvertedfromsetchesarerelatielysimple shaps，suchasfaces，humanpostures，animalpostures，shoes，andbags.Theworkbasedosketchconversionisdifultto handle images withcomplex texturestructures，makingitchallnging tosolve thetask ofimageconversionforclothing sketches. To solve this problem，anapplicationresearchonsketch toclothing image conversion is proposed basedontheabove work. Through the reverse generation of clothing simulation sketches using the -Net network model， theconstructionof the clothing sketch datasetiscompleted.Then，theimprovedBicycleGANnetwork isusedfortrainingtoachieve sketchconversion.Finally theexperimentaleffectsbeforeandaftertheimprovementarecompared.Theexperimentshowsthat theimprovedBicycleGAN network modelhassignificantimprovementcomparedtobefore，andwiththeimprovednetwork proposed inthis paper，thetask ofconverting clothing sketches to real clothing images can be effectively completed.

Keywords： image conversion; clothing sketch; Generative Adversarial Networks

0 引言

目前草圖圖像轉(zhuǎn)換的對象為人臉、人體姿態(tài)、動物姿態(tài)等較為簡單的線條，但以線條較為復(fù)雜的服裝圖像作為圖像轉(zhuǎn)換對象的應(yīng)用比較罕見。（剩余4971字）