一种改进的基于KNN颜色线性模型抠图算法
|
摘要
介绍了一种减少用户标记和改进的基于KNN(K Nearest Neighbors)颜色线性模型的图像软抠取算法。通过ESCG(Efficient Spectral Clustering on Graphs)算法对输入图像进行谱聚类,用户只需选择某些类中确定的前景、背景像素,便能生成只包含少数未知像素的三分图。基于KNN颜色线性模型的抠图算法将局部平滑假设与非局部原理相结合,但在毛发及前景背景像素近似区域抠取效果并不理想,提出的改进算法将焦点特征添加到特征向量中,最小化基于图拉普拉斯矩阵的二次目标函数并确定未知像素的透明度值。实验表明,改进算法在毛发、孔洞或者图像前景背景近似的区域都能有好的抠取效果。
An improved KNN(K Nearest Neighbors)-based color line model algorithm for efficiently extracting alphamattes and reducing users' marks is presented in this paper. It performs spectral clustering by Efficient SpectralClustering on Graphs algorithm so that users only need to select pixels which belong to definitely foreground andbackground,then it enables to generate trimap that only contains small portion of unknown pixels. KNN-based colorline model matting algorithm combines color line model and nonlocal principle,it performs poorly when imagecontains hairs,furs,and similar foreground and background regions. An improved KNN matting algorithm is proposed.The proposed algorithm adds focus information into feature vector,takes advantage of local smoothness and nonlocalprinciple,then optimization of unknown pixels' alpha by minimizing the quadratic object function based on matteLaplacian. The experiments show the improved method performs well in image scenes contain hairs,furs,holes,similarforeground and background regions.
An improved KNN(K Nearest Neighbors)-based color line model algorithm for efficiently extracting alphamattes and reducing users' marks is presented in this paper. It performs spectral clustering by Efficient SpectralClustering on Graphs algorithm so that users only need to select pixels which belong to definitely foreground andbackground,then it enables to generate trimap that only contains small portion of unknown pixels. KNN-based colorline model matting algorithm combines color line model and nonlocal principle,it performs poorly when imagecontains hairs,furs,and similar foreground and background regions. An improved KNN matting algorithm is proposed.The proposed algorithm adds focus information into feature vector,takes advantage of local smoothness and nonlocalprinciple,then optimization of unknown pixels' alpha by minimizing the quadratic object function based on matteLaplacian. The experiments show the improved method performs well in image scenes contain hairs,furs,holes,similarforeground and background regions.
引文
[1]于晏平.基于能量和取样的图像抠取算法研究[D].北京:北京交通大学,2012.
[2]关宇东,韩媞.复杂背景下基于分割逼近法的抠像技术研究[J].电视技术,2007,31(7):75-76.
[3]吕巨建,战荫伟.一种改进的Bayes抠图算法[J].计算机工程,2010,36(3):213-214.
[4]SHAHRIAN E,RAJAN D.Weighted color and texture sample selection for iamge matting[C]//Proc.2012 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2012:718-725.
[5]黄睿,王翔.改进的自然图像鲁棒抠图算法[J].计算机工程与应用,2013,49(12):136-139.
[6]LEE P,WU Y.Nonlocal matting[C]//Proc.2011 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2011:2193-2200.
[7]CHEN Q F,LI D,TANG C K.KNN matting[J].IEEE Trans.Pattern Analysis and Machine Intelligence,2013,35(9):2175-2188.
[8]LEVIN A,LISCHINSKI D,WEISS Y.A closed form solution to natural image matting[J].IEEE Trans.Pattern Analysis and Machine Intelligence,2008,30(2):228-242.
[9]SUN J,JIA J,TANG C K.Poisson matting[J].ACM Trans.Graphics,2004,23(3):315-321.
[10]JIN M,KIM B K,SONG W J.KNN-based color line model for image matting[C]//Proc.2013 IEEE International Conference on Image Processing.Melbourne:IEEE Press,2013:2480-2483.
[11]高正伟.图像前景提取技术研究[D].杭州:浙江大学,2008.
[12]彭宏京,陈松灿,张道强.一种基于局部学习的自然图像景物提取方法[J].软件学报,2009,20(4):834-844.
[13]LIU J,WANG C,DANILEVSKY M,et al.Large-scale spectral clustering on graphs[C]//Proc.23rd International Joint Conference on Artificial Intelligence.Barcelona:AAAI Press,2013:1486-1492.
[14]SUBBARAO M,CHOI T S,NIKZAD A.Focusing techniques[J].Optical Engineering,1993,32(11):2824-2836.
[15]SHI Y,AU O C,PANG J,et al.Color clustering matting[C]//Proc.2013 IEEE International Conference on Mutimedia and Expo(ICME).[S.l.]:IEEE Press,2013:1-6.
[16]MUJA M,LOWE D G.Fast approximate nearest neighbors with automatic algorithm configuration[C]//Proc.4th International Conference on Computer Vision Theory and Applications.[S.l]:INSTICC Press,2009:331-340.
[17]RHEMANN C,ROTHER C,WANG J,et al.A perceptually motivated online benchmark for image matting[C]//Proc.2009 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2009:1826-1833.
[2]关宇东,韩媞.复杂背景下基于分割逼近法的抠像技术研究[J].电视技术,2007,31(7):75-76.
[3]吕巨建,战荫伟.一种改进的Bayes抠图算法[J].计算机工程,2010,36(3):213-214.
[4]SHAHRIAN E,RAJAN D.Weighted color and texture sample selection for iamge matting[C]//Proc.2012 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2012:718-725.
[5]黄睿,王翔.改进的自然图像鲁棒抠图算法[J].计算机工程与应用,2013,49(12):136-139.
[6]LEE P,WU Y.Nonlocal matting[C]//Proc.2011 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2011:2193-2200.
[7]CHEN Q F,LI D,TANG C K.KNN matting[J].IEEE Trans.Pattern Analysis and Machine Intelligence,2013,35(9):2175-2188.
[8]LEVIN A,LISCHINSKI D,WEISS Y.A closed form solution to natural image matting[J].IEEE Trans.Pattern Analysis and Machine Intelligence,2008,30(2):228-242.
[9]SUN J,JIA J,TANG C K.Poisson matting[J].ACM Trans.Graphics,2004,23(3):315-321.
[10]JIN M,KIM B K,SONG W J.KNN-based color line model for image matting[C]//Proc.2013 IEEE International Conference on Image Processing.Melbourne:IEEE Press,2013:2480-2483.
[11]高正伟.图像前景提取技术研究[D].杭州:浙江大学,2008.
[12]彭宏京,陈松灿,张道强.一种基于局部学习的自然图像景物提取方法[J].软件学报,2009,20(4):834-844.
[13]LIU J,WANG C,DANILEVSKY M,et al.Large-scale spectral clustering on graphs[C]//Proc.23rd International Joint Conference on Artificial Intelligence.Barcelona:AAAI Press,2013:1486-1492.
[14]SUBBARAO M,CHOI T S,NIKZAD A.Focusing techniques[J].Optical Engineering,1993,32(11):2824-2836.
[15]SHI Y,AU O C,PANG J,et al.Color clustering matting[C]//Proc.2013 IEEE International Conference on Mutimedia and Expo(ICME).[S.l.]:IEEE Press,2013:1-6.
[16]MUJA M,LOWE D G.Fast approximate nearest neighbors with automatic algorithm configuration[C]//Proc.4th International Conference on Computer Vision Theory and Applications.[S.l]:INSTICC Press,2009:331-340.
[17]RHEMANN C,ROTHER C,WANG J,et al.A perceptually motivated online benchmark for image matting[C]//Proc.2009 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2009:1826-1833.