資源描述:
《基于hu不變矩的紅外圖像電力設(shè)備識別》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在工程資料-天天文庫。
1、第A-PDFSplitDEMO:Purchasefromwww.A-PDF.comtoremovethewatermark30卷第1期機電工程Vol.30No.12013年1月JournalofMechanical&ElectricalEngineeringJan.2013DOI:10.3969/j.issn.1001-4551.2013.01.002基于Hu不變矩的紅外圖像電力設(shè)備識別*11*222陳俊佑,金立軍,段紹輝,姚森敬,趙靈(1.同濟大學(xué)電子與信息工程學(xué)院,上海201804;2.深圳供電局有限公司,
2、廣東深圳518010)摘要:針對電力系統(tǒng)紅外故障檢測中對電力設(shè)備的人工識別效率低、實時性差等問題,提出了根據(jù)紅外熱圖溫度信息獲取獨立的電力設(shè)備圖像,采用計算機圖像處理技術(shù)實現(xiàn)對電力設(shè)備高效、準(zhǔn)確識別的方法。首先,通過紅外圖像中的溫度信息尋找設(shè)備中高溫點作為種子點,采用區(qū)域生長方法有效地去除了背景,獲得了整個電力設(shè)備的二值圖像;然后,選取Hu不變矩作為圖像特征提取方法,并對其做出了改進,計算了該二值圖像的Hu不變矩,構(gòu)成了電力設(shè)備的特征向量;最后,設(shè)計了BP神經(jīng)網(wǎng)絡(luò)分類器做分類識別,可用于結(jié)合溫度信息實現(xiàn)電力系統(tǒng)
3、中電力設(shè)備紅外圖像的故障識別。研究結(jié)果表明,該電力設(shè)備識別方法對CT、變壓器、母線接頭、避雷針將軍帽等電力設(shè)備的識別率高、耗時少,具有良好的應(yīng)用前景。關(guān)鍵詞:電力設(shè)備識別;紅外圖像;Hu不變矩;區(qū)域生長;BP神經(jīng)網(wǎng)絡(luò)中圖分類號:TM76;TP39文獻標(biāo)志碼:A文章編號:1001-4551(2013)01-0005-04PowerequipmentidentificationininfraredimagebasedonHuinvariantmoments11222CHENJun-you,JINLi-jun,DUA
4、NShao-hui,YAOSen-jing,ZHAOLing(1.CollegeofElectronicandInformationEngineering,TongjiUniversity,Shanghai201804,China;2.ShenzhenPowerSupplyCo.,Ltd.,Shenzhen518010,China)Abstract:Aimingattheproblemsoflowefficiencyandpoorreal-timebymanualofpowerequipmentrecognit
5、ioninpowersysteminfrareddetection,digitalimageprocessingwasproposedtorealizetheefficientandaccuraterecognitionbasedontheimagesobtainedfromtheInfraredimage.Firstly,thehightemperaturepointwasfoundasseedinpowerequipmentfromthemessageofinfraredtemperature.Thebac
6、kgroundwasremovedeffectivelybyregiongrowingmethodtoobtainthebinaryimageofentireequipment.Secondly,Huinvariantmomentsanditsimprovedalgorithmwereselectedasthemethodsoffeatureextraction.Huinvariantmomentsofbinaryimageswerecalculated,andfeaturevectorsofpowerequi
7、pmentwereobtained.Finally,classifierbasedonBPneuralnetworkwasdesignedtoachievedifferentpowerequipmentrecognition,whichwillbeusedinfaultdiagnosiswithtemperaturemessage.Theresearchresultsindicatethat,thismethodcanreceiveahighrecognitionratefordifferentequipmen
8、tandhaslesstime-consuming,sothatitwillgetagoodprospect.Keywords:powerequipmentidentification;infraredimage;Huinvariantmoments;regiongrowing;BPneuralnetwork障。熱缺陷是由于設(shè)備損壞產(chǎn)生的局部過熱,是電力0引言故障中最常見的一種,熱缺陷