Change detection(CD)is becoming indispensable for unmanned aerial vehicles(UAVs),especially in the domain of water landing,rescue and search.However,even the most advanced models require large amounts of data for model training and testing.Therefore,sufficient labeled images with different imaging conditions are needed.Inspired by computer graphics,we present a cloning method to simulate inland-water scene and collect an auto-labeled simulated dataset.The simulated dataset consists of six challenges to test the effects of dynamic background,weather,and noise on change detection models.Then,we propose an image translation framework that translates simulated images to synthetic images.This framework uses shared parameters(encoder and generator)and 22×22 receptive fields(discriminator)to generate realistic synthetic images as model training sets.The experimental results indicate that:1)different imaging challenges affect the performance of change detection models;2)compared with simulated images,synthetic images can effectively improve the accuracy of supervised models.

Peng Cheng Laboratory,Shenzhen 518000,China;State Key Laboratory of Virtual Reality Technology and Systems,School of Automation Science and Electrical Engineering,Beihang University,Beijing 100083,;Department of Automation,University of Science and Technology of China,Hefei 230027;State Key Laboratory for Management and Control of Complex Systems,Institute of Automation,Chinese Academy of Sciences,Beijing 100190,China;State Key Laboratory for Management and Control of Complex Systems,Institute of Automation,Chinese Academy of Sciences,Beijing 100190;Research Center for Computational Experiments and Parallel Systems Technology,National University of Defense Technology,Changsha 410073,China

[1]Xuan Li;Haibin Duan;Yonglin Tian;Fei-Yue Wang-.Exploring Image Generation for UAV Change Detection)[J].自动化学报（英文版）,2022(06):1061-1072

Exploring,Image,Generation,Change,Detection,detection,CD,becoming,indispensable,unmanned,aerial,vehicles,UAVs,especially,domain,water,landing,rescue,search,However,even,most,advanced,models,require,large,amounts,training,testing,Therefore,sufficient,labeled,images,different,imaging,conditions,needed,Inspired,by,computer,graphics,present,cloning,method,inland,scene,collect,auto,simulated,dataset,consists,six,challenges,effects,dynamic,background,weather,noise,change,Then,propose,translation,framework,that,translates,synthetic,This,uses,shared,parameters,encoder,generator,receptive,fields,discriminator,generate,realistic,sets,experimental,results,indicate,affect,performance,compared,can,effectively,improve,accuracy,supervised

0.649056