基于生成對(duì)抗網(wǎng)絡(luò)的服裝草圖轉(zhuǎn)換研究

打開(kāi)文本圖片集

中圖分類號(hào)：TP391.4；TP18 文獻(xiàn)標(biāo)識(shí)碼：A 文章編號(hào)：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(duì)象為人臉、人體姿態(tài)、動(dòng)物姿態(tài)等較為簡(jiǎn)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(duì)象的應(yīng)用比較罕見(jiàn)。（剩余4971字）