采用MaskR-CNN模型的城市光污染评估研究

打开文本图片集

文章编号：1674-6139（2026）04-0183-06

中图分类号：X826文献标志码：B

Urban Light Pollution Assessment Using Mask R-CNN Modeling

Wang Ju

（Shanghai Luoman Lighting Technologies Inc.，Shanghai 2Ooooo，China）

Abstract：Duringtheprocessofurbanization，theextensiveapplicationofhigh-reflectivitymaterialshsexacerbatedthepoblemof lightpolutioaditioalessmntetdsediulttoandlethsuperipodctsofpltionsouesndcompletical factors，resultingininacurateesultsTherefore，thispperpropsesanurbanlightpoltionassessmentodelbasedonmasregion convolutional neural network．Theexperimental results show that the average accuracy of ULPA model is up to 91.34% ，and the error valuesare4.8，4.nd2.8，espctively，ichisbviouslyetertanerompaisonmodelsInpracticalapcatios，UA modelcaefectielytetmpoalandspatialracteristicsoflightpliniretgios，providingsrongspfor curateaesmentandtreatmentfurbanlightpolltionTheresearchprovidesmoreaccuratedatasupportanddecisionmakingbasisfor urbanlightplltionessment，ndpromotestheplicationofeplaingtechologintefieldofurbanenvirontalgov ernance.

Keywords：glasscurtainwall；lightpolution；maskregionconvolutioalneuralnetwork；dplearningtechnology；urbanenviron mental governance

前言

随着城市发展，玻璃幕墙、釉面砖墙、磨光大理石等高反射率材料被广泛应用于建筑外立面。（剩余6714字）