Page 117 - 2024年第55卷第9期
P. 117

[38] YINH,GONGY,QIUG.Sidewindowguidedfiltering[J].SignalProcessing,2019,165:315 - 330.
                [39] LIUH,LID,DINGY.MHA - ConvLSTM damdeformationpredictionmodelconsideringenvironmentalvolumelag
                       effect[J].AppliedSciences,2023,13(14):8538.
                [40] 马怀发,王立涛,陈厚群,等.混凝土动态损伤的滞后特性[J].水利学报,2010,41(6):659 - 664.
                [41] 郑佳艳.基于序列图像的结构全域时空动态位移监测方法探索[D].重庆:重庆交通大学,2022.
                [42] 于起峰,赵宏伟,刘肖林.基于图像的精密测量与运动测量[M].北京:科学出版社,2002.



                             Machinevision - basednon - contactmonitoringmethodforgateand
                                                dam surfacedisplacements

                                                           1,2
                                                                        1,2
                                                                                    1,3
                                   CHENBo   1,2,3 ,HEMengjia ,LIUWeiqi ,MACong
                            (1.StateKeyLaboratoryofWaterDisasterPrevention,HohaiUniversity,Nanjing 210098,China;
                          2.CollegeofWaterConservancyandHydropowerEngineering,HohaiUniversity,Nanjing 210098,China;
                                 3.InstituteofWaterEngineeringSafety,HohaiUniversity,Nanjing 210098,China)

                  Abstract:Aimingattheproblemsofconventionalmonitoringmethodsuchashighlabordensity,lowmonitoring
                  frequency ,anddifficultyinachievinglong - termstablemonitoringofthesurfacedisplacementoflocksanddams,
                  anon - contactintelligentmonitoringmethodintegratingspatio - temporalfeaturesisproposed.Themethodadoptsan
                  artificialtargetasamarker ,takesacameraasanacquisitiondevice,transmitsimageinformationwirelessly,and
                  makesuseofAdaptiveGammaCorrectionWeightedDistribution(AGCWD)andWeightedGuidedImageFilter
                  (WGIF)withimprovededge - awarenessfactor.WGIFwithedge - awarenessfactortoenhancethefeatureexpression
                  abilityoflowilluminationimages ,andtheSpatio - TemporalContext(STC)algorithmbasedonBayesianframework
                  onboardthecomputertodeeplyminethecontextualspatio - temporalinformationofthetargetimage ,andfurther
                  introducesurfacefittingtoobtainsub - pixelleveldisplacementinformationofthetargettoachievethesub - pixel
                  displacementinformationofbothhorizontalandverticalbi - directionalsurfacedisplacementsofthegateanddam.
                  Further ,surfacefittingisintroducedtoobtainsubpixel - leveldisplacementinformationtoachievenon - pixel - level
                  non - contactmonitoringofhorizontalandverticalbi - directionalsurfacedisplacementsofthelockanddam.Thela
                  boratoryandfieldtestresultsshowthatthedisplacementmonitoringdataunderdifferentexperimentalscenariosare
                  highlyconsistentwiththecalibrationdata,andtheerrorislessthan0.05mm;comparedwiththeexistingimage
                  optimizationprocessingmethods ,theimageoptimizationprocessingmethodsbasedonAGCWDandWGIFincrease
                  thePeakSignal - to - NoiseRatio (PSNR)by2.70%,increasetheinformationentropyby4.91%,andreducethe
                  standarddeviationby2.63%;comparedwiththeestablishedtargettrackingalgorithms,theSTCtargettrackingal
                  gorithmbasedonsurfacefittingismoreeffectiveinobtainingsub - pixel - leveldisplacementinformation.Compared
                  withtheexistingtargettrackingalgorithms,thefieldmonitoringdataoftheSTCtargettrackingalgorithmbasedon
                  surfacefittingimprovestheaccuracyofthesimilartargettrackingalgorithmsby48%,whichcanprovideahigh -
                  precisionsolutionforthemonitoringofthegateanddamsurfacedisplacements.
                  Keywords:lockanddam;surfacedisplacementmonitoring;spatio - temporalcharacteristicsofimagesequences;
                  digitalimageoptimisation;sub - pixelleveltargettracking

                                                                                    (责任编辑:王 婧)

















                —  1 1 2 —
                     2
   112   113   114   115   116   117   118   119   120   121   122