遥感数据处理中多源数据融合方法研究

打开文本图片集

中图分类号：P237 文献标志码：A 文章编号：2095-2945(2025)14-0162-04

1,2,3（1.,西安710001；2.陕西省第三测绘工程院,西安71001；3.西安航空学院,西安 710089）

Abstract:Inremotesensing dataprocesing，multi-sourcedata fusion method hasbecomea keytechnologytoimprove the eficiencyandaccuracyof informationextraction.Thispapersystematicalldiscussesthemulti-sourcedatafusionmethodbased onfeaturespaceanddeeplearning,analyzesthediferencebetweenlinearandnonlinearfusionanditsapplicationintheproce offeaturextractionandselection,andfurtherelaboratestheimportanceoffeaturespacedimensionreductiontechnology.Dep learningtechniques，speciallconvolutionalneuralnetworksandself-supervisedlearning，havedemonstratedexcellent performanceinprocessngheterogeneousandmulti-dimensionalremotesensingdata，signficantlyimprovingtheaccuracyand robustnessofdatafusion.Basedonpracticalcases,thispapershowsthespecificefectsofdiferentfusionmethodsonfeature extractionandfusioninmulti-sourcedataprocessng，indicatingthatdeeplearning methodshavebroadapplicationprospectsin the field of remote sensing.

Keywords: remote sensing data; multi-source data fusion; feature space; fusion effect; deep learning

在遥感技术的不断进步与广泛应用背景下，多源数据融合已成为提升数据处理精度与丰富信息内容的关键手段，遥感数据来源广泛，涵盖光学、雷达、激光雷达等多种传感器，这些数据在空间、时间及光谱分辨率上各具特性。（剩余5629字）