鲁棒自适应加权的引导滤波算法
|
摘要
为了提高图像滤波时边缘的保持能力,提出鲁棒自适应加权的引导滤波算法。首先利用一阶差分法判断高斯滤波处理后引导图像的边缘位置信息,在去除噪声干扰的同时,提高边缘信息提取的鲁棒性,然后通过最大类间方差法(Otsu)分割边缘区域与非边缘区域,提高区域阈值选取的自适应性,最后利用改进的分段函数模型拟合理想权重因子,控制不同区域的平滑程度,实现鲁棒自适应引导滤波,达到保边平滑的目的。通过图像平滑实验与抠图实验对所提算法性能进行了验证,与引导滤波算法及另外2种改进算法相比,所提算法的保边平滑性能更强。
To improve the edge-preserving ability of image filtering a guided image filtering algorithm based on robust adaptive weighting is proposed. Firstly Gaussian filter is performed on the guided image and firstorder differential method is used to judge the position of the edge which can remove the noise as well as improve the robustness of the edge information extraction. Then Otsu is used to segment the edge regions from the non-edge regions the self-adaptability of threshold selection is improved. Finally the improved piecewise function model is used to fit the ideal weight factor and control the degree of smoothness of different regions which can realize the robust adaptive guided filtering and achieve the smoothness of the edge. The experiments of edge-preserving smoothing and matting are carried out. Compared with guided image filtering algorithm and the other two improved algorithms the proposed algorithm has better robustness and edge-preserving smoothness.
To improve the edge-preserving ability of image filtering a guided image filtering algorithm based on robust adaptive weighting is proposed. Firstly Gaussian filter is performed on the guided image and firstorder differential method is used to judge the position of the edge which can remove the noise as well as improve the robustness of the edge information extraction. Then Otsu is used to segment the edge regions from the non-edge regions the self-adaptability of threshold selection is improved. Finally the improved piecewise function model is used to fit the ideal weight factor and control the degree of smoothness of different regions which can realize the robust adaptive guided filtering and achieve the smoothness of the edge. The experiments of edge-preserving smoothing and matting are carried out. Compared with guided image filtering algorithm and the other two improved algorithms the proposed algorithm has better robustness and edge-preserving smoothness.
引文
[1] KO S JYONG H L. Center weighted median filters and their applications to image enhancement[J]. IEEE Transactions on Circuits&Systems199138(9):984-993.
[2] PERONA PMALIK J. Scale-space and edge detection using anisotropic diffusion[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence199012(7):629-639.
[3] BUADES ACOLL BMOREL J M. The staircasing effect in neighborhood filters and its solution[J]. IEEE Transactions on Image Processing200615(6):1499-1505.
[4] TOMASI CMANDUCHI R. Bilateral filtering for gray and color images[C]//Proceedings of the 6th International Conference on Computer VisionIEEE1998:839-846.
[5] HE K MSUN JTANG X O. Guided image filtering[J].IEEE Transactions on Pattern Analysis&Machine Intelligence201335(6):1397-1409.
[6] LI Z GZHENG J HZHU Z Jet al. Weighted guided image filtering[J]. IEEE Transactions on Image Processing201524(1):120-129.
[7]龙鹏,鲁华祥. Lo G边缘算子改进的加权引导滤波算法[J].计算机应用,2015,35(9):2661-2665.
[8]谢伟,周玉钦,游敏.融合梯度信息的改进引导滤波[J].中国图象图形学报,2016,21(9):1119-1126.
[9] OTSU N. A threshold selection method from gray-level histograms[J]. IEEE Transactions on SystemsMan&Cybernetics19799(1):62-66.
[10] WANG ZBOVIK A CSHEIKH H Ret al. Image quality assessment:from error visibility to structural similarity[J]. IEEE Transactions on Image Processing200413(4):600-612.
[11] NARVEKAR N DKARAM L J. A no-reference perceptual image sharpness metric based on a cumulative probability of blur detection[C]//International Workshop on Quality of Multimedia ExperienceIEEE2009:87-91.
[2] PERONA PMALIK J. Scale-space and edge detection using anisotropic diffusion[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence199012(7):629-639.
[3] BUADES ACOLL BMOREL J M. The staircasing effect in neighborhood filters and its solution[J]. IEEE Transactions on Image Processing200615(6):1499-1505.
[4] TOMASI CMANDUCHI R. Bilateral filtering for gray and color images[C]//Proceedings of the 6th International Conference on Computer VisionIEEE1998:839-846.
[5] HE K MSUN JTANG X O. Guided image filtering[J].IEEE Transactions on Pattern Analysis&Machine Intelligence201335(6):1397-1409.
[6] LI Z GZHENG J HZHU Z Jet al. Weighted guided image filtering[J]. IEEE Transactions on Image Processing201524(1):120-129.
[7]龙鹏,鲁华祥. Lo G边缘算子改进的加权引导滤波算法[J].计算机应用,2015,35(9):2661-2665.
[8]谢伟,周玉钦,游敏.融合梯度信息的改进引导滤波[J].中国图象图形学报,2016,21(9):1119-1126.
[9] OTSU N. A threshold selection method from gray-level histograms[J]. IEEE Transactions on SystemsMan&Cybernetics19799(1):62-66.
[10] WANG ZBOVIK A CSHEIKH H Ret al. Image quality assessment:from error visibility to structural similarity[J]. IEEE Transactions on Image Processing200413(4):600-612.
[11] NARVEKAR N DKARAM L J. A no-reference perceptual image sharpness metric based on a cumulative probability of blur detection[C]//International Workshop on Quality of Multimedia ExperienceIEEE2009:87-91.