基于邊緣特征增強的油菜田塊信息提取方法

打開文本圖片集

Abstract：Obtaining high-quality features is akeystep in extracting high-precision oilsed rape plot information from remote sensing images.To address theproblemof theunsatisfactoryqualityof theedgefeaturesofoilseedrapefields，this paperproposesanEdgeFeature EnhancementNet （EFENet）foroilseedrapefieldinformation extractionmodel.Firstly，the encoder-decoder structureisadoptedasthe basicstructureof thefeatureextractor，and theEdge Atention Mechanism （EAM） is designed as the feature enhancement module，comprising two sub-modules：the channel atention and spatial atention，toimprovethefeaturequalityoftheedgepixels.Secondly，theConsideringBoundaryEnhancementLossFunction （CBELoss）isdesigned toimprove theedge featurequalityconsistingofa weightterm @and Mathematical Expectationof Characteristic Mass（MECM）. ω expresss the efect of sample location on the error based on the pixel spatialcorrelation， and the MECMenhanced the discriminative degree of features by evaluating the quality of the samples. EFENet employs SoftMax asapixel bypixel classifier.Inthispaper，PMS（Gaofen-6Panchromaticand MultispectralScanner）imagesare selectedas thedatasources，Menyuan Hui AutonomousCounty，Haibei TibetanAutonomousPrefectureandQinghai Province as the studyarea，ERFNet，RefineNet，and UNetas thecomparison models.Theresults show that the proposed method outperforms the comparison model in terms of F1 score （92.40%） ，recall rate （93.64%） ，checking accuracy （92.83%） ， and precision （92.51%） ），indicating thatthe model has obvious advantages in extracting the informationofGF-6PMSoilseed rape fields.

Keywords：Convolutionalneuralnetwork;oilseedrapefieldinformation;edgefeatureenhancement;lossfunction;feature quality

糧油產業(yè)已成為青海省經濟發(fā)展的支柱產業(yè)和農民增收的重要渠道，油菜作為青海省第一大油料作物，占全省油料作物播種面積的1/2，是青海省糧油生產的重要支柱[。（剩余13130字）