面向復(fù)雜場(chǎng)景目標(biāo)提取的顏色增強(qiáng)與連通區(qū)域構(gòu)造方法
Color enhancement and connected region construction method for object extraction in complex scenes
LIMinglu’,ZHAO Chunqing2,HOUMaoxin2,WANGXiaoxia',YANGFengbao1 (1.School of Informationand Communication Engineering,North Universityof China,Taiyuan O3oo51,China; 2.Colectivellbed
Abstract:Itisdificulttoeffectivelyextractthecompleteobjectregionincomplexscenesbytheexistingfeatureextraction methods,soamethodofcolor enhancementandconnectedregion construction basedonHSVspacenonlinear isproposed.On thebasisoftheobjectcolorfeatures,thehistogramdistributioncharacteristicsareanalyzedtoobtainthemultilevelthresholdsfor objectextraction.Forthecaseofporobjectextractionduetodiferentlightingconditions,theenancedextractionisperfored byconstructingthenonlinear mappingrelationshipbetweenhueandluminanceunderdiferentlightingconditions.Fortheregion of poorconnectivityintheextractedimage,thenearest-neighborskeletonconnection methodisproposedtoenhance the connectivityoftheobjectregion,soastomakeitconstituteacompleteobjectcontour.Bythedetectioncomparisonexperiments ofmulti-sceneobject extraction,theoverallaveragedetectionaccuracyrateof theproposed methodisimprovedbyO.1026in comparisonwiththatof themorphological processingmethd.Therefore,theobjectextractionmethodproposedcanimprovethe object detection accuracy rate of theunmanned detection platformsin complex scenes.
Keywords:colorspace;threshold segmentation;color enhancement;object extraction;skeleton extraction;narest-neighbor connection algorithm
0 引言
無(wú)人探測(cè)平臺(tái)因其高靈活性、綜合多種傳感器的探測(cè)能力以及實(shí)時(shí)通信作業(yè)等優(yōu)勢(shì),為智能檢測(cè)技術(shù)的發(fā)展提供了有力支撐。(剩余9512字)
-
-
- 現(xiàn)代電子技術(shù)
- 2025年13期
- 基于FNM-Net的輕量級(jí)遙感...
- 基于YOLOv8n的輕量化道路...
- 面向復(fù)雜場(chǎng)景目標(biāo)提取的顏色增強(qiáng)...
- 基于改進(jìn)U-Net的細(xì)胞核圖像...
- 基于深度學(xué)習(xí)和Retinex理...
- 基于注意力機(jī)制和ACT網(wǎng)絡(luò)的人...
- 基于改進(jìn)RT-DETR的小目標(biāo)...
- 基于級(jí)聯(lián)式逆殘差網(wǎng)絡(luò)的游戲圖像...
- 基于顯著性特征的多視角動(dòng)作圖像...
- 新能源接人下移動(dòng)通信傳輸網(wǎng)絡(luò)控...
- 基于FMCW毫米波雷達(dá)遠(yuǎn)于2m...
- 基于樣本重要性的分布式深度學(xué)習(xí)...
- 基于RDMMIMOOFDM雷達(dá)...
- T-BOI:一種融合時(shí)間和行為...
- 基于改進(jìn)的灰狼算法優(yōu)化BP神經(jīng)...
- 基于改進(jìn)的ResNet網(wǎng)絡(luò)和特...
- 基于深度特征融合的惡意軟件檢測(cè)...
- 融合雙通道特征信息的醫(yī)療短文本...
- 聲源定位系統(tǒng)的廣義二次互相關(guān)算...
- 基于GAIL方法的魚(yú)類個(gè)體運(yùn)動(dòng)...
- 基于圖像知識(shí)增強(qiáng)的中文多模態(tài)反...
- 基于ZYNQ-7000和AD9...
- 基于自適應(yīng)采樣的全息圖像壓縮感...
- 基于電感電容的鋰離子電池組雙層...
- 基于獨(dú)立線長(zhǎng)預(yù)測(cè)信息的低功耗驅(qū)...
- 基于YOLOv8的多功能導(dǎo)盲系...