基于复杂背景下绝缘子的识别

摘要由于输电线路长期在恶劣的自然环境下运行，加上长期机械负荷和强电场的因素，会使绝缘子产生故障，所以要对输电线路进行定期巡检。然而因为架设的输电线路需要经过山川河流等许多的复杂地形，人工检测的难度非常大。当前，输电线路的巡检已经普遍采用直升机实行数据采集，但是怎样从冗杂的数据中检测出输电线路中各部件的故障是目前研究工作的要点。本篇论文的主要目的是研究航拍输电线路图像中绝缘子部件的分割与识别，这也正是绝缘子故障检测的基础工作，并且可以直接用于直升机航拍巡检图像中绝缘子部件的提取。该方法首先将RGB色彩空间分为三个单独通道，并选用B通道进行分割，然后采用sobel算子边缘检测法对图像进行分割处理，接着用形态学滤波的方法滤除噪声，再用连通区域标记的方法将分割出的绝缘子串进行标记，最后采用基于面积的特征提取方法完整地将绝缘子提取出。经过实验表明，本文的方法能够有效的识别并提取出目标绝缘子图像，但是由于航拍图像中复杂的背景干扰以及航拍角度等外在因素的干扰，就会给目标的提取带来一定的困难，所以还需要对提取方法加以研究和改进。关键词：数字图像处理，图像去噪，绝缘子提取，连通区域标记，形态学滤波，特征提取ABSTRACTDuetothetransmissionlinerunningforalongtimeinyourbadnaturalenvironment,andfactorssuchasthemechanicalloadandstrongelectricfieldforalongtime,canmakethemalfunctionoftheinsulatorsoneedregularcheckingontransmissionlines.Butduetothelayingofthetransmissionlinepassesthroughmountainsandriversandmanyothercomplexterrain,andthedifficultyofmanualinspection.Atpresent,theinspectionofthetransmissionlinehasbeenwidelyusedinhelicopterfordatacollection,andhowtodetectfrommultifariousdataforthepartsofthetransmissionlinefaultisthefocusofcurrentresearch.Inthispaper,themainworkistostudytheaerialtransmissionlineinsulatorinimagesegmentationandrecognition,itisalsothefoundationoffaultidentification,canbeusedfortheextractionofimagefromahelicopteraerialinspectioninsulatorimages.Thismethodfirstofall,theRGBcolorspaceisdividedintothreeseparatechannels,andselectstheBchannelsegmentation.Andthenapplysobeledgeoperatormethodforimagesegmentation,then,withmorphologicalfilteringmethodtofilteroutnoise.Tousethemethodoftheconnectedareawillsplitouttheinsulatorstring,finallyadoptsthemethodoffeatureextractioniscompletetopickouttheinsulator.AfterHalconexperimentsimulationsoftware,themethodcaneffectivelyfromcomplexbackgroundimagesfullofextractingtargetimage.Saidbyexperiment,themethodcaneffectivelyextractthetargets,duetothecomplexbackgroundofaerialimagesandaerialperspective,andaseriesofexternalfactors,canbringcertaindifficultytotheextractionoftarget,sostillneedtoimproveonextractionmethod.KEYWORDS:digitalimageprocessing,imagedenoising,Insulatorextract,ConnectedComponentLabeling,morphologicalfilter,featureextraction目录摘要...............................................................................................................................IABSTRACT..................................................................................................................II第1章绪论................................................................................................................11.1课题研究背景及意义.......................................................................................11.2国内外研究现状...............................................................................................11.3本课题研究工作与结构安排...........................................................................3第2章RGB色彩空间...................................................................................................52.1RGB彩色空间介绍..........................................................................................5第3章图像分割..........................................................................................................73.1图像分割简介...................................................................................................73.2图像分割方法概述...........................................................................................73.2.1基于阈值的分割方法............................................................................73.2.2基于区域的图像分割............................................................................93.3.3基于边缘检测的图像分割..................................................................103.3实验结果对比分析.........................................................................................12第4章形态学处理....................................................................................................154.1形态学处理简介.............................................................................................154.2形态学处理方法概述.....................................................................................154.2.1腐蚀：..................................................................................................164.2.2膨胀......................................................................................................164.2.3开运算..................................................................................................164.2.4闭运算..................................................................................................174.3实验结果对比分析.........................................................................................17第5章连通区域标记................................................................................................205.1连通区域标记简介.........................................................................................205.2连通区域标记方法概述.................................................................................205.2.1象素标记..............................................................................................205.2.2游程连通性分析.................................................