基于Sobel算子的Criminisi改进算法
|
摘要
对于大尺寸破损图像的修复,Criminisi算法以其在修补结果和修补速度方面的优势得到广泛应用。但是Cri-minisi算法在优先权函数的定义上存在缺陷,其计算结果难以对图像中的纹理部分和边缘部分进行很好地辨别,从而会使修补结果中出现纹理延伸现象,破坏图像的原始结构。针对这一现象,论文引入Sobel边缘检测算子改进优先权函数,在原有的优先权计算基础上引入边缘项,并且利用权重因子改变边缘项在优先权计算中所占的比重,提升了算法对边缘的辨识能力,有效抑制了修补结果中的纹理延伸现象,通过仿真结果表明改进算法优于原有的Criminisi算法。
The Criminisi algorithm has been widely used with its advantage in the inpainting results and the inpainting speed of the large size damaged image inpainting. But the Criminisi algorithm has defect in the definition of the priority function which is not very good sense of texture and the edge part of the images,it will make the inpainting results in texture extension phenomenon. In view of this phenomenon,this paper introduces an improved priority function based on Sobel operator which introduces the edge part on the basis of the original algorithm and uses the weight factor for a change in the proportion of priority calculation to enhance the identification ability of the algorithm to edge and effectively inhibit the texture extension phenomenon. According to the simulation results,the improved algorithm is better than the Criminisi algorithm.
The Criminisi algorithm has been widely used with its advantage in the inpainting results and the inpainting speed of the large size damaged image inpainting. But the Criminisi algorithm has defect in the definition of the priority function which is not very good sense of texture and the edge part of the images,it will make the inpainting results in texture extension phenomenon. In view of this phenomenon,this paper introduces an improved priority function based on Sobel operator which introduces the edge part on the basis of the original algorithm and uses the weight factor for a change in the proportion of priority calculation to enhance the identification ability of the algorithm to edge and effectively inhibit the texture extension phenomenon. According to the simulation results,the improved algorithm is better than the Criminisi algorithm.
引文
[1]张红英,彭启琮.数字图像修复技术综述[J].中国图象图形学报,2007,12(1):1-10.ZHANG Hongying,PENG Qicong.A Survey on Digital Image Inpainting[J].Journal of Image and Graphics,2007,12(1):1-10.
[2]M.Bertalmio,G.Sapiro,C.Ballester,and V.Caselles.Image inpainting[J].Siggraph,2005:417-424.
[3]T.F.Chan,J.Shen.Mathematical models for local nontexture inpaintings[J].SIAM J.Appl.Math,2001,62(3):1019-1043.
[4]H.Y.Zhang,B.Wu,Q.C.Peng,Y.D.Wu.Digitalimage inpainting based on p-harmonic energy minimization[J].Chin.J.Electron,2007,3(3):525-530.
[5]A.Criminisi,P.Pérez,and K.Toyama.Region fillingand object removal by exemplar-based inpainting[J].IEEET-rans.Image Process,2004,13(9):1200-1212.
[6]任澍,唐向宏,康佳伦.纹理和边缘特征相结合的图像修补算法[J].计算机辅助设计与图形学学报,2013,25(11):1682-1692.REN Shu,TANG Xianghong,KANG Jialun.An Image Inpainting Algorithm Combined with Texture and Edge Features[J].Journal of Computer-Aided Design and Computer Graphics,2013,25(11):1682-1692.
[7]J.Wu,Q.Ruan.Object removal by cross isophotesexemplar based image inpainting[J].Proceeding of International Conference of Pattern Recognition,2006:810-813.
[8]Z.Xu,J.Sun.Image inpainting by patch propagation using patch sparsity[J].IEEE Trans.Image Process,2010,19(5):1153-1165.
[9]O.Le Meur,J.Gautier,C.Guillemot.Examplar-based inpainting based on local geometry[J].in Proc.IEEEInt.Conf.Image Processing(ICIP),2011:3401-3403.
[10]许凯,王智,阎石,等.基于结构相似度的图像修补算法[J].计算机时代,2015,10:7-10.XU Kai,WANG zhi,YAN shi,et al.Image inpainting algorithm based on structuralsimilarity[J].ComputerEra,2015,10:7-10.
[11]何春华,张雪飞,胡迎春.基于改进Sobel算子的边缘检测算法的研究[J].光学技术,2012,38(3):323-327.HE Chunhua,ZHANG Xuefei,HU Yingchun.A study on the improved algorithm for Sobel on image edge detection[J].Optical Technique,2012,38(3):323-327.
[2]M.Bertalmio,G.Sapiro,C.Ballester,and V.Caselles.Image inpainting[J].Siggraph,2005:417-424.
[3]T.F.Chan,J.Shen.Mathematical models for local nontexture inpaintings[J].SIAM J.Appl.Math,2001,62(3):1019-1043.
[4]H.Y.Zhang,B.Wu,Q.C.Peng,Y.D.Wu.Digitalimage inpainting based on p-harmonic energy minimization[J].Chin.J.Electron,2007,3(3):525-530.
[5]A.Criminisi,P.Pérez,and K.Toyama.Region fillingand object removal by exemplar-based inpainting[J].IEEET-rans.Image Process,2004,13(9):1200-1212.
[6]任澍,唐向宏,康佳伦.纹理和边缘特征相结合的图像修补算法[J].计算机辅助设计与图形学学报,2013,25(11):1682-1692.REN Shu,TANG Xianghong,KANG Jialun.An Image Inpainting Algorithm Combined with Texture and Edge Features[J].Journal of Computer-Aided Design and Computer Graphics,2013,25(11):1682-1692.
[7]J.Wu,Q.Ruan.Object removal by cross isophotesexemplar based image inpainting[J].Proceeding of International Conference of Pattern Recognition,2006:810-813.
[8]Z.Xu,J.Sun.Image inpainting by patch propagation using patch sparsity[J].IEEE Trans.Image Process,2010,19(5):1153-1165.
[9]O.Le Meur,J.Gautier,C.Guillemot.Examplar-based inpainting based on local geometry[J].in Proc.IEEEInt.Conf.Image Processing(ICIP),2011:3401-3403.
[10]许凯,王智,阎石,等.基于结构相似度的图像修补算法[J].计算机时代,2015,10:7-10.XU Kai,WANG zhi,YAN shi,et al.Image inpainting algorithm based on structuralsimilarity[J].ComputerEra,2015,10:7-10.
[11]何春华,张雪飞,胡迎春.基于改进Sobel算子的边缘检测算法的研究[J].光学技术,2012,38(3):323-327.HE Chunhua,ZHANG Xuefei,HU Yingchun.A study on the improved algorithm for Sobel on image edge detection[J].Optical Technique,2012,38(3):323-327.