遙感數(shù)據(jù)處理中多源數(shù)據(jù)融合方法研究

打開(kāi)文本圖片集

中圖分類號(hào)：P237 文獻(xiàn)標(biāo)志碼：A 文章編號(hào)：2095-2945(2025)14-0162-04

1,2,3（1.,西安710001；2.陜西省第三測(cè)繪工程院,西安71001；3.西安航空學(xué)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

在遙感技術(shù)的不斷進(jìn)步與廣泛應(yīng)用背景下，多源數(shù)據(jù)融合已成為提升數(shù)據(jù)處理精度與豐富信息內(nèi)容的關(guān)鍵手段，遙感數(shù)據(jù)來(lái)源廣泛，涵蓋光學(xué)、雷達(dá)、激光雷達(dá)等多種傳感器，這些數(shù)據(jù)在空間、時(shí)間及光譜分辨率上各具特性。（剩余5629字）