摘要
保持结构的图像滤波是诸多计算摄影学和图像分析应用的必要预处理方法.传统的纹理滤波方法仅仅考虑了图像的局部对比信息,当图像中包含对比度较强的纹理时,这些纹理会干扰现有方法,导致边界模糊等副作用.为此,提出一种基于块漂移的联合非局部均值滤波的方法.首先利用块漂移的方法选择出最能表示每个像素点纹理信息的图像块,生成具有判别性的引导图,该引导图在抑制纹理信息的同时保持住了原图的结构部分;然后借助该引导图进行非局部的联合滤波,通过考察图像块之间的相似性进一步降低由于引导图的模糊而引起的滤波后图像边界的模糊现象,从而得到清晰的图像结构信息.与其他经典方法进行实验的结果表明,该方法可以更好地移除纹理并且能够保持结构信息;此外,还将纹理移除应用到边缘检测和纹理增强,显示了保持结构的纹理移除有助于提高二者的效果.
Texture filtering with structure-preservation is a primary pre-processing method in many applications of computational photography and image analysis. Conventional methods may fail and generate blurred edges when the input image contains textures with strong contrast, since they just consider local contrast information of an image. Therefore, we propose a non-local joint texture removing method based on patch shift. First, a guidance map which preserves edges while suppressing texture is acquired via patch shift. Then a non-local joint texture filtering is applied with the assistance of the guidance map. By utilizing the similarity between non-local patches, we further suppress the blur caused by the guidance map and extract distinct image structures. We show that our method outperforms previous methods in removing texture while preserving primary image structures through a large number of experiments. Moreover, we perform the edge detection and detail enhancement on the images after texture removal and illustrate that our method benefits such application scenarios.
引文
[1]Wei L Y,Lefebvre S,Kwatra V,et al.State of the art in example-based texture synthesis[C]//Proceedings of Eurographics’s State of the Art Report.Aire-la-Ville:Eurographics Association Press,2009:93-117
[2]Farbman Z,Fattal R,Lischinski D,et al.Edge-preserving decompositions for multi-scale tone and detail manipulation[J].ACM Transactions on Graphics,2008,27(3):Article No.67
[3]Kartic S,Cyril S,Durand F.Edge-preserving multiscale image decomposition based on local extrema[J].ACM Transactions on Graphics,2009,28(5):Article No.147
[4]Xu L,Yan Q,Xia Y,et al.Structure extraction from texture via relative total variation[J].ACM Transactions on Graphics,2012,31(6):Article No.139
[5]Li Bo,Su Zhuo,Leng Chengcai,et al.Gradient minimized Mumford-Shah model for high-dimensional filtering[J].Acta Automatic Sinica,2014,40(12):2926-2935(in Chinese)(李波,苏卓,冷成财,等.基于混合梯度最小化Mumford-Shah模型的高维滤波算法[J].自动化学报,2014,40(12):2926-2935)
[6]Tomasi C,Manduchi R.Bilateral filtering for gray and color images[C]//Proceedings of the 6th IEEE International Conference on Computer Vision.Washington D C:IEEE Computer Society Press,1998:839-846
[7]Cho H,Lee H,Kang H,et al.Bilateral texture filtering[J].ACM Transactions on Graphics,2014,33(4):Article No.128
[8]Zhang Q,Shen X Y,Xu L,et al.Rolling guidance filter[M]//Lecture Notes in Computer Science.Heidelberg:Springer,2014,8691:815-830
[9]Buades A,Coll B,Morel JM.A non-local algorithm for image denoising[C]//Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition.Washington D C:IEEE Computer Society Press,2005,2:60-65
[10]Canny J.A computational approach to edge detection[J].IEEETransactions on Pattern Analysis&Machine Intelligence,1986,8(6):679-698
[11]Li Ming,Yang Yanping.TV-Retinex:a fast algorithm for image enhancement[J].Journal of Computer-Aided Design&Computer Graphics,2010,22(10):1777-1782(in Chinese)(李明,杨艳屏.TV-Retinex:一种快速图像增强算法[J].计算机辅助设计与图形学学报,2010,22(10):1777-1782)
