
中圖分類號:TN911.73-34;TP391 文獻標識碼:A 文章編號:1004-373X(2025)16-0123-05
Visual defocusing image dark edge local restoration algorithm for considering highand lowfrequencyfeatures
LIU Xiao
(XinyangNormal University,Xinyang464Ooo, China)
Abstract:Thedarkedgesofvisualdefocusingimagesareblurred,resulting inthelossofdetailinformation,whichis manifestedinthereductionofhighandlowfrequencycomponentsintheimage tovaryingdegrees.Therefore,avisual defocusingimagesdarkedgeslocalrestorationalgorithmconsideringhighandlowfrequencyfeaturesisstudiedtorestorethe edgedetailsofsuchimages.Adefocusingimagemodelisconstructed,andtheehancementprocessingofthevisualdefocusing imageinthemodelisconductedbymeansof thepalalgorithm,soastoobtaintheenhancedgraydefocusingimage.Thenonsubsampledcontour wave transform (NSCT)algorithm isusedtodecomposetheenhanceddefocusing image with multi-scaleand multi-direction,soastoextractthehighandlowfrequencyfeaturesinformationsuchascompleteedgeandstructureoftheimage. ByimplementingNSCTinversetransformationonsuchhighandlowfrequencyfeatureinformation,thehighandlowfrequency edgeimagesareobtained,andthelowfrequencyedgeimagesarecompensatedintothehigh frequencyedgeimage,soastobtain thefinalreconstructededgerestorationimage.Theresultsshowthatthealgorithmcanachievelocalrestorationofthedarkedge ofdifferentscaledefocusingimages.Afterrestoration,theedgedetailsareclearandcomplete,thestructuralinformationis recovered well,the brightnessandcontrast are improved obviously,andtheoverall restoration performance is stable.
Keywords:highandlow frequency feature;visual defocusing image;dark edge;local restoration;imageenhancement; grayscale image;NSCTalgorithm
0 引言
在圖像實際拍攝過程中,由于拍攝設備的質量問題、調焦不準、手部抖動等各種原因,圖像往往會產生視覺離焦模糊現象,特別是圖像的邊緣部分會因亮度較低產生暗邊緣,導致圖像細節丟失,這會嚴重影響圖像的后續應用3。暗邊緣局部復原對于提高此類圖像整體質量、恢復圖像的視覺效果具有重要意義[4-5]。
目前,已有部分中外學者針對此領域展開了相關研究。例如,文獻[6]研究的離焦圖像重建算法通過改進去噪自編碼器提取離焦圖像特征,融合后輸人生成對抗網絡內,獲得重建圖像。但該方法在復原離焦圖像時,無法完全恢復圖像中的細節信息。文獻[7]提出的圖像重建算法運用構建的圖像退化模型預處理原始退化圖像,結合卷積神經網絡(CNN)提取處理后圖像特征并獲得其重建圖像。CNN主要依賴于局部感受野和卷積核實現圖像特征提取,在處理需提取廣泛區域特征的離焦圖像時,可能導致部分細節信息丟失,復原圖像不準確。文獻[8]研究的圖像復原算法結合空間域卷積設計實時圖像復原算法,將該算法映射到FPGA上,獲得復原圖像。盡管FPGA具有高速并行處理能力,但離焦圖像復原需要高精度的計算及精細的圖像處理技術,才能恢復圖像中的細節和紋理,FPGA在這方面存在一定的局限性。文獻[9]研究的離焦模糊圖像復原算法運用點擴散函數設計最小二乘算法,通過該算法獲得原始離焦圖像的復原圖像,對此類復原圖像進行去振鈴處理后,得到最終的復原圖像。……