We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
基于生成对抗网络的遥感影像色彩一致性方法.
- Authors
王艺儒; 王光辉; 杨化超; 刘慧杰
- Abstract
Urightness emd colors are p+rne to occur 1.310, emd Uetwee. captured images i. the process of remote sensing imaaing. However, the manual color conditioning combined with image processing soOwarr con no longer meet the color matching demand of geometricolly increesing remote sensing images. Given this, this study proposed a kind of unsupervised channee - cycle generative adversarial network (CA - CycleGAN) inWgrated with the attention mechanism suitabk fco ground objects in complex urban areas with a high land utilization rate. Firstly, the sample data set used foe coIov referenco was manuHy made through histogem adjustwent and Photoshop, and eie appropvate urban irndges were selected as eie sample sS te be corrected. Then, the two kinds of inidges were cut respectively t obtain the preprocessed irndge sample sets. Findlly, Wc preprocessed nnaae set te be corrected and the nnaae set Ov coIov referenco were processed using the CA 一 CycleGAN. Beceuse the atention mechanism has been added te the generatOT, the gemated focuses cen be distrinuted ine key arees using the attention feature map in the training process of the confrontation between tee geneeatoeand thedneoenmnnatoe, thuenmpeoenngthenmageeeoteand obtannnngtheooeoeooeeotnon modeebaeed on urban nnages and We images aftea coIov cooection. Both the imaae cooection effed and the loss function diaaram show that the proposed method is optimized based on the CycleGAN and that the comprehensive performance of We CyoeGANnntgeatd wnth th,atntnon m,ohannemneb,tethan thatwnthoutth,atntnon m,ohannem. Compaed to oone,ntnonaem,thode, th,m,thod peopoe,d nn thneetudygeateyeduo,d th,tnm,ooeooeoeooe,otnon and aohne,d more stable nnage color correction tfects than manual color matching. Therefore, the method proposed in this study eneoy//ngnnonoantadeantage/nn theooeoedodgnngooeemote/en/nngnmage/and ha/agood appenoatnon peo/peot
- Subjects
GENERATIVE adversarial networks; REMOTE sensing; IMAGE processing; LAND use; CITIES &; towns
- Publication
Remote Sensing for Natural Resources, 2022, Vol 34, Issue 3, p65
- ISSN
2097-034X
- Publication type
Article
- DOI
10.6046/zrzyyg.2021316