基于特征圖的路面凍融裂縫紅外檢測(cè)算法

打開文本圖片集

中圖分類號(hào)：U416.2 文獻(xiàn)標(biāo)志碼：A 文章編號(hào)：1005-8249（2025）02-0164-05

DOI：10.19860/j.cnki.issn1005-8249.2025.02.028

Abstract：Aimingatthe problem of porconvergenceeffectoffreeze-thaw crack characteristicsandlowaccuracyof pavement freeze-thaw crack detection，an infrared detection algorithm for pavement freeze-thaw cracks basedon feature maps was designed.Basedontheprincipleofinfraredthermal imaging，anewmethodforreconstructing thetemperaturefieldofpavement freze-thawcrack infrared thermal imagesisformedbycombininganisotropictemperaturefieldreconstructionandbilateral flteringteory.Afreze-thawcrackfeature matrix isconstructed，andthepavementfreze-thawcrack infrared iage isused asinputtooptimizetheconvolutionkernelweightsoftheconvolutionalneuralnetwork．Afeaturemapbaseddetectionmethodis constructedtoobtainthebestpavementfreze-thawcrackinfraredimagingdetectionresults.Theresultsshowthatthealgorithm can maketheinfrared images ofroad freeze-thaw cracksclearerafter temperaturefieldreconstruction，and preservetheedge informationandcontouroftheimage.Moreover，underdiferenttypesof noise interference，thereconstructed temperature field imagealwayshasahighpeaksignal-to-noiseratio.This methodhasdetecteddiferent typesof pavementfreeze-thawcracks， which can improve the effectiveness of road safety operation and maintenance

Key words：road freeze-thawcracks；convolutionalneural network；infraredimaging；thermalimage denoising；crack detection； machine learning

0 引言

隨著我國(guó)基礎(chǔ)設(shè)施建設(shè)的快速推進(jìn)，對(duì)路面結(jié)構(gòu)耐久性提出了更高要求。（剩余4893字）