資源描述:
《基于邊緣特征的匹配算法改進研究》由會員上傳分享,免費在線閱讀,更多相關內容在學術論文-天天文庫。
1、頌.1:論義某于邊緣特征的匹配算法改進研究摘要針對圖像匹配全局搜索影響匹配速度的問題,本文引入了遺傳算法,將其與Hausdom距離相結合來優(yōu)化圖像匹配的搜索過程。首先對基準圖像和實時圖像進行預處理,以自適應選取閡值的Canny算子來提取圖像的邊緣作為特征單元;然后引入遺傳算法對待配準圖像進行匹配操作;在遺傳操作中利用改進的STMHausdorff足巨離來構造遺傳算法的適應度函數(shù),以此通過遺傳算法確定出最優(yōu)的變換參數(shù),完成基準圖像和實時圖像的匹配;最后通過實驗仿真驗證了算法的有效性。Hausdor艘巨離作為一種評價兩個圖像位置關系的量化標準,以其很強的抗干擾和容錯能力
2、被廣泛應用于圖像匹配之中。然而單純的Hausdorff足巨離對噪聲和孤立點均比較敏感,導致誤匹配率較高。各類改進的Hausdorff足巨離形式能在特定的匹配環(huán)境中克服這些不足,但面對本課題中復雜的成像畸變和復合噪聲的情況,仍不能獲得理想的匹配效果。本文采取了一種先對標準方差的Hausdorfi膃巨離(sTMHD)進行排序,爾后再求其部分均值的Hausdo硪距離改進形式。這種經改進的ST心D能較好的克服噪聲、偽邊緣及部分遮擋對匹配精度和穩(wěn)定性的影響,在配準速度和精度上取得較為理想的效果。關鍵詞:圖像匹配,邊緣,c鋤ly算子,Hausdor唧巨離,遺傳算法Abstrac
3、t頌I:論文Globalsearchfortheimagematchingthespeedoftheimpactofmatchingproblems,thispaper,weintroduceageneticalgorithm,withtheHausdorffdistancewillbeacombinationofimagematchingtooptimizethesearchprocess.First,thereferenceimagesandreal—timeimagepre—processing,inordertoselectadaptivethreshold
4、Cannyoperatortoextracttheedgeimageasafeatureunit.Andthendealwiththeintroductionofgeneticalgorithmtomatchtheimageregistrationoperation.GeneticmanipulationintheuseofstandarddeviationmodifiedHausdorffdistanceimprovedgeneticalgorithmtoconstructthefitnessfunction,geneticalgorithmtodetermine
5、theoptimaltransformationparameters,thecompletionofthereferenceimagesandreal-timeimagematching.Finally,throughexperimentsimulationtheeffectivenessofthealgorithm.●Hausdorffdistancebetweentwoimagesasanevaluationoftherelationshipbetweenthelocationofthequantitativecriteria,、加thitsstronganti
6、—interferenceabilityandfaulttolerancearewidelyusedinimagematching.Hausdorffdistance,howeversimpleonthenoiseandisolatedpointsaremoresensitive,resultinginahigherrateoffalsematches.VarioustypesofimprovementintheformofHausdorffdistancematchingaspecificenvironmenttoovercomethesedeficiencies
7、,butinthefaceofthecomplexityoftheissueofdistortionoftheimagingnoiseandcomplexsituation,theyCallnotachieveanidealeffectmatch.TlliSarticletakesafirststandarddeviationmodifiedHausdorffdistance(STMHD)sort,fortheirpartofSeoulafterthemeanimprovementsintheformofHausdorffdistance.STMHDthisim