《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (2): 560-566.DOI: 10.11772/j.issn.1001-9081.2021122168
所属专题: 多媒体计算与计算机仿真
收稿日期:
2021-12-29
修回日期:
2022-04-14
接受日期:
2022-04-19
发布日期:
2022-05-16
出版日期:
2023-02-10
通讯作者:
孙忠贵
作者简介:
孙梦迪(1997—),女,山东滨州人,硕士研究生,主要研究方向:图像处理、机器学习基金资助:
Mengdi SUN, Zhonggui SUN(), Xu KONG, Hongyan HAN
Received:
2021-12-29
Revised:
2022-04-14
Accepted:
2022-04-19
Online:
2022-05-16
Published:
2023-02-10
Contact:
Zhonggui SUN
About author:
SUN Mengdi, born in 1997, M. S. candidate. Her research interests include image processing, machine learning.Supported by:
摘要:
针对传统数学形态学(TMM)细节保持能力较差,以及现有自适应改进方法数学性质丢失的问题,提出了一种针对多模态图像的自适应引导形态学(GAMM)。首先,通过考虑输入图像和引导图像的联合信息进行结构元素的构建,从而在一定程度上增强了相应算子对噪声的鲁棒性;其次,借助3σ原则,使结构元素成员的选取能够自适应于图像内容;最后,利用稀疏矩阵的哈达玛积对结构元素施加一个对称性约束。理论证明和仿真实验均表明所提形态学的相应算子能够同时具备保序性和附益性等重要数学性质。在多模态图像上进行去噪实验,结果表明GAMM比TMM以及近年所提出的鲁棒自适应形态学(RAMM)在峰值信噪比(PSNR)上高出约2~3 dB;同时,主观视觉效果对比表明了GAMM在噪声去除、结构保持方面明显优于TMM和RAMM。
中图分类号:
孙梦迪, 孙忠贵, 孔旭, 韩红燕. 针对多模态图像的自适应引导形态学设计[J]. 计算机应用, 2023, 43(2): 560-566.
Mengdi SUN, Zhonggui SUN, Xu KONG, Hongyan HAN. Design of guided adaptive mathematical morphology for multimodal images[J]. Journal of Computer Applications, 2023, 43(2): 560-566.
图像 | 滤波窗口 | 输入 | TMM | RAMM | GAMM | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
18.66 | 14.55 | 30.59 | 28.14 | 24.80 | 29.38 | 27.63 | 24.65 | 32.70 | 28.23 | 24.81 | |||
28.77 | 27.75 | 25.91 | 28.87 | 27.60 | 25.93 | 31.21 | 27.86 | 25.93 | |||||
27.37 | 27.07 | 25.51 | 27.57 | 27.12 | 25.68 | 30.33 | 27.57 | 25.73 | |||||
Book | 18.63 | 14.53 | 31.60 | 27.63 | 24.64 | 31.41 | 27.54 | 24.65 | 33.66 | 27.63 | 24.67 | ||
30.53 | 27.70 | 26.20 | 30.47 | 27.65 | 26.03 | 33.66 | 27.75 | 26.29 | |||||
30.11 | 27.57 | 26.37 | 30.26 | 27.58 | 26.42 | 33.64 | 27.96 | 26.42 | |||||
Doll | 18.64 | 14.53 | 32.65 | 27.93 | 25.00 | 32.35 | 27.91 | 25.05 | 33.87 | 27.94 | 25.95 | ||
31.98 | 28.68 | 26.78 | 32.06 | 28.64 | 26.67 | 34.19 | 28.69 | 26.79 | |||||
31.62 | 28.79 | 26.86 | 31.67 | 28.79 | 27.04 | 34.26 | 28.92 | 27.96 | |||||
Laundry | 14.48 | 33.90 | 28.95 | 25.35 | 33.21 | 28.83 | 25.31 | 34.07 | 29.01 | 25.41 | |||
31.91 | 29.78 | 26.67 | 32.02 | 29.71 | 26.64 | 33.62 | 29.82 | 26.83 | |||||
31.41 | 29.33 | 26.04 | 31.42 | 29.46 | 26.21 | 32.59 | 29.84 | 26.36 | |||||
Moebius | 14.47 | 31.80 | 27.98 | 24.91 | 31.42 | 27.79 | 24.91 | 33.44 | 28.08 | 24.93 | |||
30.40 | 28.32 | 26.32 | 30.48 | 28.20 | 26.24 | 32.91 | 28.82 | 26.55 | |||||
29.94 | 24.97 | 26.36 | 30.02 | 27.91 | 26.46 | 32.79 | 28.12 | 27.10 | |||||
Reindeer | 18.65 | 14.57 | 33.46 | 28.74 | 25.12 | 30.95 | 28.10 | 24.96 | 33.93 | 28.81 | 25.18 | ||
29.98 | 29.24 | 26.98 | 30.26 | 28.99 | 26.75 | 33.57 | 30.01 | 27.86 | |||||
28.00 | 29.02 | 26.95 | 28.20 | 28.78 | 26.95 | 30.83 | 29.26 | 26.95 |
表1 彩色图像-深度图像去噪的PSNR结果 (dB)
Tab. 1 Results of PSNR in RGB-depth image denoising
图像 | 滤波窗口 | 输入 | TMM | RAMM | GAMM | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
18.66 | 14.55 | 30.59 | 28.14 | 24.80 | 29.38 | 27.63 | 24.65 | 32.70 | 28.23 | 24.81 | |||
28.77 | 27.75 | 25.91 | 28.87 | 27.60 | 25.93 | 31.21 | 27.86 | 25.93 | |||||
27.37 | 27.07 | 25.51 | 27.57 | 27.12 | 25.68 | 30.33 | 27.57 | 25.73 | |||||
Book | 18.63 | 14.53 | 31.60 | 27.63 | 24.64 | 31.41 | 27.54 | 24.65 | 33.66 | 27.63 | 24.67 | ||
30.53 | 27.70 | 26.20 | 30.47 | 27.65 | 26.03 | 33.66 | 27.75 | 26.29 | |||||
30.11 | 27.57 | 26.37 | 30.26 | 27.58 | 26.42 | 33.64 | 27.96 | 26.42 | |||||
Doll | 18.64 | 14.53 | 32.65 | 27.93 | 25.00 | 32.35 | 27.91 | 25.05 | 33.87 | 27.94 | 25.95 | ||
31.98 | 28.68 | 26.78 | 32.06 | 28.64 | 26.67 | 34.19 | 28.69 | 26.79 | |||||
31.62 | 28.79 | 26.86 | 31.67 | 28.79 | 27.04 | 34.26 | 28.92 | 27.96 | |||||
Laundry | 14.48 | 33.90 | 28.95 | 25.35 | 33.21 | 28.83 | 25.31 | 34.07 | 29.01 | 25.41 | |||
31.91 | 29.78 | 26.67 | 32.02 | 29.71 | 26.64 | 33.62 | 29.82 | 26.83 | |||||
31.41 | 29.33 | 26.04 | 31.42 | 29.46 | 26.21 | 32.59 | 29.84 | 26.36 | |||||
Moebius | 14.47 | 31.80 | 27.98 | 24.91 | 31.42 | 27.79 | 24.91 | 33.44 | 28.08 | 24.93 | |||
30.40 | 28.32 | 26.32 | 30.48 | 28.20 | 26.24 | 32.91 | 28.82 | 26.55 | |||||
29.94 | 24.97 | 26.36 | 30.02 | 27.91 | 26.46 | 32.79 | 28.12 | 27.10 | |||||
Reindeer | 18.65 | 14.57 | 33.46 | 28.74 | 25.12 | 30.95 | 28.10 | 24.96 | 33.93 | 28.81 | 25.18 | ||
29.98 | 29.24 | 26.98 | 30.26 | 28.99 | 26.75 | 33.57 | 30.01 | 27.86 | |||||
28.00 | 29.02 | 26.95 | 28.20 | 28.78 | 26.95 | 30.83 | 29.26 | 26.95 |
图像 | 滤波窗口 | 输入 | TMM | RAMM | GAMM | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0.155 | 0.080 | 0.891 | 0.678 | 0.473 | 0.871 | 0.664 | 0.468 | 0.896 | 0.678 | 0.481 | |||
0.871 | 0.801 | 0.678 | 0.872 | 0.795 | 0.665 | 0.873 | 0.857 | 0.679 | |||||
0.845 | 0.821 | 0.651 | 0.844 | 0.817 | 0.639 | 0.846 | 0.821 | 0.691 | |||||
Book | 0.103 | 0.050 | 0.906 | 0.656 | 0.439 | 0.900 | 0.649 | 0.432 | 0.934 | 0.657 | 0.439 | ||
0.924 | 0.818 | 0.700 | 0.923 | 0.807 | 0.676 | 0.973 | 0.872 | 0.754 | |||||
0.924 | 0.871 | 0.805 | 0.926 | 0.865 | 0.786 | 0.951 | 0.874 | 0.836 | |||||
Doll | 0.484 | 0.106 | 0.048 | 0.885 | 0.650 | 0.445 | 0.878 | 0.644 | 0.442 | 0.895 | 0.655 | 0.446 | |
0.893 | 0.806 | 0.686 | 0.900 | 0.797 | 0.666 | 0.900 | 0.815 | 0.692 | |||||
0.887 | 0.854 | 0.789 | 0.888 | 0.848 | 0.775 | 0.895 | 0.891 | 0.797 | |||||
Laundry | 0.480 | 0.050 | 0.911 | 0.662 | 0.461 | 0.902 | 0.658 | 0.454 | 0.946 | 0.662 | 0.471 | ||
0.923 | 0.825 | 0.697 | 0.920 | 0.815 | 0.678 | 0.972 | 0.869 | 0.751 | |||||
0.918 | 0.865 | 0.776 | 0.916 | 0.862 | 0.767 | 0.949 | 0.890 | 0.775 | |||||
Moebius | 0.482 | 0.050 | 0.898 | 0.659 | 0.449 | 0.891 | 0.652 | 0.443 | 0.943 | 0.659 | 0.453 | ||
0.909 | 0.815 | 0.688 | 0.910 | 0.806 | 0.666 | 0.969 | 0.865 | 0.696 | |||||
0.900 | 0.858 | 0.777 | 0.901 | 0.853 | 0.766 | 0.949 | 0.887 | 0.782 | |||||
Reindeer | 0.497 | 0.113 | 0.057 | 0.918 | 0.669 | 0.458 | 0.905 | 0.661 | 0.450 | 0.951 | 0.675 | 0.459 | |
0.920 | 0.826 | 0.704 | 0.922 | 0.814 | 0.685 | 0.981 | 0.882 | 0.767 | |||||
0.900 | 0.870 | 0.807 | 0.903 | 0.863 | 0.790 | 0.953 | 0.874 | 0.842 |
表2 彩色图像-深度图像去噪的SSIM结果
Tab. 2 Results of SSIM in RGB-depth image denoising
图像 | 滤波窗口 | 输入 | TMM | RAMM | GAMM | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0.155 | 0.080 | 0.891 | 0.678 | 0.473 | 0.871 | 0.664 | 0.468 | 0.896 | 0.678 | 0.481 | |||
0.871 | 0.801 | 0.678 | 0.872 | 0.795 | 0.665 | 0.873 | 0.857 | 0.679 | |||||
0.845 | 0.821 | 0.651 | 0.844 | 0.817 | 0.639 | 0.846 | 0.821 | 0.691 | |||||
Book | 0.103 | 0.050 | 0.906 | 0.656 | 0.439 | 0.900 | 0.649 | 0.432 | 0.934 | 0.657 | 0.439 | ||
0.924 | 0.818 | 0.700 | 0.923 | 0.807 | 0.676 | 0.973 | 0.872 | 0.754 | |||||
0.924 | 0.871 | 0.805 | 0.926 | 0.865 | 0.786 | 0.951 | 0.874 | 0.836 | |||||
Doll | 0.484 | 0.106 | 0.048 | 0.885 | 0.650 | 0.445 | 0.878 | 0.644 | 0.442 | 0.895 | 0.655 | 0.446 | |
0.893 | 0.806 | 0.686 | 0.900 | 0.797 | 0.666 | 0.900 | 0.815 | 0.692 | |||||
0.887 | 0.854 | 0.789 | 0.888 | 0.848 | 0.775 | 0.895 | 0.891 | 0.797 | |||||
Laundry | 0.480 | 0.050 | 0.911 | 0.662 | 0.461 | 0.902 | 0.658 | 0.454 | 0.946 | 0.662 | 0.471 | ||
0.923 | 0.825 | 0.697 | 0.920 | 0.815 | 0.678 | 0.972 | 0.869 | 0.751 | |||||
0.918 | 0.865 | 0.776 | 0.916 | 0.862 | 0.767 | 0.949 | 0.890 | 0.775 | |||||
Moebius | 0.482 | 0.050 | 0.898 | 0.659 | 0.449 | 0.891 | 0.652 | 0.443 | 0.943 | 0.659 | 0.453 | ||
0.909 | 0.815 | 0.688 | 0.910 | 0.806 | 0.666 | 0.969 | 0.865 | 0.696 | |||||
0.900 | 0.858 | 0.777 | 0.901 | 0.853 | 0.766 | 0.949 | 0.887 | 0.782 | |||||
Reindeer | 0.497 | 0.113 | 0.057 | 0.918 | 0.669 | 0.458 | 0.905 | 0.661 | 0.450 | 0.951 | 0.675 | 0.459 | |
0.920 | 0.826 | 0.704 | 0.922 | 0.814 | 0.685 | 0.981 | 0.882 | 0.767 | |||||
0.900 | 0.870 | 0.807 | 0.903 | 0.863 | 0.790 | 0.953 | 0.874 | 0.842 |
1 | SERRA J. Image Analysis and Mathematical Morphology[M]. London: Academic Press, 1982:536. 10.2307/2531038 |
2 | LEI T, JIA X H, LIU T L, et al. Adaptive morphological reconstruction for seeded image segmentation[J]. IEEE Transactions on Image Processing, 2019, 28(11):5510-5523. 10.1109/tip.2019.2920514 |
3 | HUANG W L, LIU J X. Robust seismic image interpolation with mathematical morphological constraint[J]. IEEE Transactions on Image Processing, 2020, 29:819-829. 10.1109/tip.2019.2936744 |
4 | GONZALEZ R C, WOODS R E, EDDINS S L. Digital Image Processing Using Matlab[M]. 2nd ed. Knoxville, TN: Gatesmark Publishing, 2009:486-534. |
5 | SOILLE P. Morphological Image Analysis: Principles and Applications[M]. 2nd. Berlin: Springer, 2004. 10.1007/978-3-662-05088-0 |
6 | ROERDINK J B T M. Adaptivity and group invariance in mathematical morphology[C]// Proceedings of the 16th IEEE International Conference on Image Processing. Piscataway: IEEE, 2009:2253-2256. 10.1109/icip.2009.5413983 |
7 | VELASCO-FORERO S, ANGUIO J. On nonlocal mathematical morphology[C]// Proceedings of the 2013 International Symposium on Mathematical Morphology and Its Applications to Signal and Image Processing, LNCS 7883. Berlin: Springer, 2013:219-230. |
8 | ĆURIĆ V, LANDSTRÖM A, THURLEY M J, et al. Adaptive mathematical morphology - a survey of the field[J]. Pattern Recognition Letters, 2014, 47:18-28. 10.1016/j.patrec.2014.02.022 |
9 | DEBAYLE J, PINOLI J C. Spatially adaptive morphological image filtering using intrinsic structuring elements[J]. Image Analysis and Stereology, 2005, 24(3):145-158. |
10 | LERALLUT R, DECENCIÈRE É, MEYER F. Image filtering using morphological amoebas[J]. Image and Vision Computing, 2007, 25(4):395-404. 10.1016/j.imavis.2006.04.018 |
11 | CUISENAIRE O. Locally adaptable mathematical morphology using distance transformations[J]. Pattern Recognition, 2006, 39(3):405-416. 10.1016/j.patcog.2005.07.009 |
12 | TREECE G. The bitonic filter: linear filtering in an edge-preserving morphological framework[J]. IEEE Transactions on Image Processing, 2016, 25(11): 5199-5211. 10.1109/tip.2016.2605302 |
13 | MILANFAR P. A tour of modern image filtering: new insights and methods, both practical and theoretical[J]. IEEE Signal Processing Magazine, 2013, 30(1):106-128. 10.1109/msp.2011.2179329 |
14 | PETSCHNIGG G, SZELISKI R, AGRAWALA M, et al. Digital photography with flash and no-flash image pairs[J]. ACM Transactions on Graphics, 2004, 23(3):664-672. 10.1145/1015706.1015777 |
15 | HE K M, SUN J, TANG X O. Guided image filtering[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(6): 1397-1409. 10.1109/tpami.2012.213 |
16 | LI Z G, ZHENG J H, ZHU Z J, et al. Weighted guided image filtering[J]. IEEE Transactions on Image Processing, 2015, 24(1):120-129. 10.1109/tip.2014.2371234 |
17 | SUN Z G, HAN B, LI J, et al. Weighted guided image filtering with steering kernel[J]. IEEE Transactions on Image Processing, 2020, 29:500-508. 10.1109/tip.2019.2928631 |
18 | 韩博,孙忠贵. 基于结构相似性感知的引导图像滤波[J]. 聊城大学学报(自然科学版), 2020,33(5):9-15. |
HAN B, SUN Z G. Guided image filtering based on structure similarity aware[J]. Journal of Liaocheng University (Natural Science Edition), 2020, 33(5):9-15. | |
19 | PUKELSHEIM F. The three sigma rule[J]. American Statistician, 1994, 48(2):88-91. 10.1080/00031305.1994.10476030 |
20 | SUSSNER P. Lattice fuzzy transforms from the perspective of mathematical morphology[J]. Fuzzy Sets and Systems, 2016, 288:115-128. 10.1016/j.fss.2015.09.018 |
21 | SUN Z G, GAO X B, ZHANG D M. A nonlocal operator model for morphological image processing[C]// Proceedings of the 2017 IEEE International Conference on Image Processing. Piscataway: IEEE, 2017:230-234. 10.1109/icip.2017.8296277 |
22 | 孙忠贵,高新波,张冬梅,等. 一个具有附益性及保序性算子的形态学非局部拓展[J]. 电子学报, 2018, 46(8):1969-1975. 10.3969/j.issn.0372-2112.2018.08.023 |
SUN Z G, GAO X B, ZHANG D M, et al. A nonlocal extension of morphology with adjunction and order-preservation operators[J]. Acta Electronica Sinica, 2018, 46(8): 1969-1975. 10.3969/j.issn.0372-2112.2018.08.023 | |
23 | 王朋伟,牛瑞卿. 基于灰度形态学与小波相位滤波的高分辨率遥感影像边缘检测[J]. 计算机应用, 2011, 31(9): 2481-2484. 10.3724/SP.J.1087.2011.02481 |
WANG P W, NIU R Q. Edge detection of high resolution remote sensing images based on morphology and wavelet phase filtering[J]. Journal of Computer Applications, 2011, 31(9): 2481-2484. 10.3724/SP.J.1087.2011.02481 | |
24 | SHUI P L, WANG F P. Anti-impulse-noise edge detection via anisotropic morphological directional derivatives[J]. IEEE Transactions on Image Processing, 2017, 26(10):4962-4977. 10.1109/tip.2017.2726190 |
25 | SHUANG X C, ZHANG K. An edge detection algorithm based on adaptive weight multi-structure element mathematical morphology[J]. Energy Procedia, 2011, 13:9983-9991. |
26 | RONSE C. Why mathematical morphology needs complete lattices[J]. Signal Processing, 1990, 21(2):129-154. 10.1016/0165-1684(90)90046-2 |
27 | VERD-MONEDERO R, ANGULO J, SERRA J. Anisotropic morphological filters with spatially-variant structuring elements based on image-dependent gradient fields[J]. IEEE Transactions on Image Processing, 2011, 20(1):200-212. 10.1109/tip.2010.2056377 |
28 | APTOULA E, LEFÈVRE S. A comparative study on multivariate mathematical morphology[J]. Pattern Recognition, 2007, 40(11):2914-2929. 10.1016/j.patcog.2007.02.004 |
29 | SHEN X Y, ZHOU C, XU L, et al. Mutual-structure for joint filtering[J]. International Journal of Computer Vision, 2017, 125(1/2/3):19-33. 10.1007/s11263-017-1021-y |
30 | YAN Q, SHEN X Y, XU L, et al. Cross-field joint image restoration via scale map[C]// Proceedings of the 2013 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2013:1537-1544. 10.1109/iccv.2013.194 |
[1] | 陈学斌, 任志强, 张宏扬. 联邦学习中的安全威胁与防御措施综述[J]. 《计算机应用》唯一官方网站, 2024, 44(6): 1663-1672. |
[2] | 董炜娜, 刘佳, 潘晓中, 陈立峰, 孙文权. 基于编码-解码网络的大容量鲁棒图像隐写方案[J]. 《计算机应用》唯一官方网站, 2024, 44(3): 772-779. |
[3] | 李博, 黄建强, 黄东强, 王晓英. 基于异构平台的稀疏矩阵向量乘自适应计算优化[J]. 《计算机应用》唯一官方网站, 2024, 44(12): 3867-3875. |
[4] | 黄杰, 武瑞梓, 李均利. 高效的自适应复杂网络鲁棒性优化算法[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3530-3539. |
[5] | 涂进兴, 李志雄, 黄建强. 基于GPU对角稀疏矩阵向量乘法的动态划分算法[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3521-3529. |
[6] | 汪韩, 万源, 王东, 丁义明. 宽度学习系统中鲁棒性权值矩阵组合的筛选方法[J]. 《计算机应用》唯一官方网站, 2024, 44(10): 3032-3038. |
[7] | 赵徐炎, 崔允贺, 蒋朝惠, 钱清, 申国伟, 郭春, 李显超. CHAIN:基于重合支配的边缘计算节点放置算法[J]. 《计算机应用》唯一官方网站, 2023, 43(9): 2812-2818. |
[8] | 高健, 李智, 樊缤, 姜传贤. 基于光线投射采样和四元数正交矩的高效三维医学影像鲁棒零水印算法[J]. 《计算机应用》唯一官方网站, 2023, 43(4): 1191-1197. |
[9] | 葛孟婷, 万鸣华. 基于近邻监督局部不变鲁棒主成分分析的特征提取模型[J]. 《计算机应用》唯一官方网站, 2023, 43(4): 1013-1020. |
[10] | 范贤博俊, 陈立家, 李珅, 王晨露, 王敏, 王赞, 刘名果. 鲁棒的视觉机械臂联合建模优化方法[J]. 《计算机应用》唯一官方网站, 2023, 43(3): 962-971. |
[11] | 刘丽, 陈长波. 带状稀疏矩阵乘法及高效GPU实现[J]. 《计算机应用》唯一官方网站, 2023, 43(12): 3856-3867. |
[12] | 柏财通, 崔翛龙, 郑会吉, 李爱. 基于自监督知识迁移的鲁棒性语音识别技术[J]. 《计算机应用》唯一官方网站, 2022, 42(10): 3217-3223. |
[13] | 高工, 杨红雨, 刘洪. 基于深度学习的三维点云人脸识别[J]. 计算机应用, 2021, 41(9): 2736-2740. |
[14] | 董阳, 潘海为, 崔倩娜, 边晓菲, 滕腾, 王邦菊. 面向多模态磁共振脑瘤图像的小样本分割方法[J]. 计算机应用, 2021, 41(4): 1049-1054. |
[15] | 李华, 卢桂馥, 余沁茹. 基于干净数据的流形正则化非负矩阵分解[J]. 《计算机应用》唯一官方网站, 2021, 41(12): 3492-3498. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||