Please wait a minute...
浙江大学学报(工学版)  2024, Vol. 58 Issue (10): 2053-2061    DOI: 10.3785/j.issn.1008-973X.2024.10.008
计算机与控制工程     
基于峰值感知和多尺度约束的加权引导滤波器
张全(),刘海忠*()
兰州交通大学 数理学院,甘肃 兰州 730070
Weighted guided filter based on peak-aware and multi-scale constraints
Quan ZHANG(),Haizhong LIU*()
College of Mathematics and Physics, Lanzhou Jiaotong University, Lanzhou 730070, China
 全文: PDF(3099 KB)   HTML
摘要:

引导图像滤波无法保留锐利边缘,导致平滑图像出现结构过度模糊问题, 为此提出新的加权引导图像滤波器. 利用峰值感知加权提取图像的边缘和结构信息,通过多尺度约束提高所提滤波器的鲁棒性. 基于图像方差信息将滤波损失函数的正则化项系数改进为自适应式. 边缘感知平滑、图像细节增强、纹理去除平滑以及图像去噪领域的应用实验结果表明, 所提滤波器在视觉效果、峰值信噪比以及结构相似度上均优于参与对比的引导图像滤波器. 与边缘感知平滑实验中次优滤波器的峰值信噪比和结构相似度相比,所提滤波器的峰值信噪比平均高2.62 dB, 结构相似度平均高0.0286.

关键词: 引导滤波器图像平滑峰值感知多尺度约束边缘保持    
Abstract:

A new weighted guided image filter was proposed for the problem that guided image filtering fails to preserve sharp edges and leads to excessive structural blurring in smoothed images. Peak-aware weighting was utilized to extract edge and structural information from images, and the robustness of the proposed filter was improved by multi-scale constraints. The regularization term coefficients of the filter loss function were improved to an adaptive form based on the image variance information. Application experiments were carried out in edge-aware smoothing, image detail enhancement, texture removal smoothing, and image denoising, and the results showed that the proposed filter outperformed the guided image filters involved in the comparison in terms of visualization, peak signal-to-noise ratio, and structural similarity. Compared with the peak signal-to-noise ratio and structural similarity of the suboptimal filters of the edge-aware smoothing experiments, the peak signal-to-noise ratio was 2.62 dB higher on average, and the structural similarity was 0.0286 higher on average.

Key words: guided filter    image smoothing    peak aware    multi-scale constraint    edge preserving
收稿日期: 2023-06-16 出版日期: 2024-09-27
CLC:  TP 391  
基金资助: 甘肃省优秀研究生“创新之星”项目(2023CXZX-592).
通讯作者: 刘海忠     E-mail: 12211601@stu.lzjtu.edu.cn;liuhzh@lzjtu.edu.cn
作者简介: 张全(1996—),男,硕士生,从事图像处理研究. orcid.org/0000-0003-4382-2947. E-mail:12211601@stu.lzjtu.edu.cn
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章  
张全
刘海忠

引用本文:

张全,刘海忠. 基于峰值感知和多尺度约束的加权引导滤波器[J]. 浙江大学学报(工学版), 2024, 58(10): 2053-2061.

Quan ZHANG,Haizhong LIU. Weighted guided filter based on peak-aware and multi-scale constraints. Journal of ZheJiang University (Engineering Science), 2024, 58(10): 2053-2061.

链接本文:

https://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2024.10.008        https://www.zjujournals.com/eng/CN/Y2024/V58/I10/2053

图 1  不同滤波器的模型系数比较
$r$$\lambda $SSIM
模型1模型2模型3
20.010.90930.90260.9727
40.010.87830.88920.9675
80.010.85610.89890.9701
20.100.81260.76310.9518
40.100.72730.67020.9305
80.100.65260.60800.9194
21.000.73920.71020.9474
41.000.61440.58090.9201
81.000.49990.47140.9001
表 1  不同滤波器模型的结构相似度对比
$r$$\lambda $/$ {\lambda _0} $自适应前自适应后
PSNR/dBSSIMPSNR/dBSSIM
20.0137.670.972739.100.9763
40.0136.680.967538.340.9727
80.0136.290.970138.180.9754
20.1034.380.951835.120.9559
40.1032.150.930533.140.9382
80.1030.350.919431.730.9320
21.0033.850.947433.990.9482
41.0031.310.920131.540.9220
81.0029.070.900129.460.9045
表 2  正则化项系数自适应前后的模型性能对比
图 2  边缘感知平滑实验原图
图 3  不同滤波器模型的图像平滑结果视觉对比
$\lambda $模型PSNR/dBSSIM
$r = 2$$r = 4$$r = 8$$r = 2$$r = 4$$r = 8$
0.12GIF30.9830.1729.710.90410.89780.9076
WGIF31.2930.4229.920.90630.89960.9092
GGIF36.0334.9934.570.94950.94340.9485
SWGIF37.3234.4732.430.98300.96410.9473
SKWGIF32.2431.2230.790.91990.90470.9135
本研究39.1038.3438.180.97630.97540.9627
0.22GIF26.8525.2623.900.81050.76490.7506
WGIF27.2625.5924.140.81740.77210.7571
GGIF34.1432.0530.470.92910.90710.8111
SWGIF34.3830.3127.440.97310.92700.8758
SKWGIF28.4626.5525.110.85180.79090.7718
本研究36.2334.6433.680.96280.95110.9502
0.42GIF24.5322.5320.670.73680.64740.5860
WGIF24.9122.8120.840.74510.65600.5934
GGIF33.5130.8028.390.92080.88710.8663
SWGIF32.9628.0224.580.96770.89610.8011
SKWGIF26.2823.8321.780.79990.69210.6210
本研究34.7332.6031.010.95330.93300.9239
表 3  不同滤波器模型的图像平滑结果对比
图 4  输入图像和6个模型滤波输出的第180个水平剖面的像素强度分布
图 5  不同滤波器模型的图像细节增强效果视觉对比
图 6  不同滤波器模型的纹理去除效果视觉对比
图 7  去噪测试用图像
模型PSNR/dBSSIMTS/s
GIF27.140.71500.006
WGIF28.510.74610.006
GGIF32.520.86830.009
SWGIF28.260.75300.048
SKWGIF27.350.689722.795
本研究34.710.92440.010
表 4  不同滤波器模型的图像去噪结果对比
图 8  去噪可视化实验用图像
图 9  不同滤波器模型的图像去噪结果视觉对比
20 MAGNIER B, MONTESINOS P, DIEP D. Texture removal by pixel classification using a rotating filter [C]// 2011 IEEE International Conference on Acoustics, Speech and Signal Processing . Prague: IEEE, 2011: 1097–1100.
21 HE K, SUN J, TANG X Guided image filtering[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35 (6): 1397- 1409
doi: 10.1109/TPAMI.2012.213
22 LI Z, ZHENG J, ZHU Z, et al Weighted guided image filtering[J]. IEEE Transactions on Image Processing, 2015, 24 (1): 120- 129
doi: 10.1109/TIP.2014.2371234
23 KOU F, CHEN W, WEN C, et al Gradient domain guided image filtering[J]. IEEE Transactions on Image Processing, 2015, 24 (11): 4528- 4539
doi: 10.1109/TIP.2015.2468183
24 YIN H, GONG Y, QIU G Side window guided filtering[J]. Signal Processing, 2019, 165: 315- 330
doi: 10.1016/j.sigpro.2019.07.026
25 SUN Z, HAN B, LI J, et al Weighted guided image filtering with steering kernel[J]. IEEE Transactions on Image Processing, 2020, 29: 500- 508
doi: 10.1109/TIP.2019.2928631
26 DENG G, GALETTO F, Al-NASRAWI M, et al A guided edge-aware smoothing-sharpening filter based on patch interpolation model and generalized gamma distribution[J]. IEEE Open Journal of Signal Processing, 2021, 2: 119- 135
doi: 10.1109/OJSP.2021.3063076
27 LIU D, HE C Peak-aware guided filtering for spectrum signal denoising[J]. Chemometrics and Intelligent Laboratory Systems, 2022, 222: 104508
doi: 10.1016/j.chemolab.2022.104508
28 LI Y, LI Z, ZENG C, et al Adaptive weighted guided image filtering for depth enhancement in shape-from-focus[J]. Pattern Recognition, 2022, 131: 108900
doi: 10.1016/j.patcog.2022.108900
29 XU H, ZHANG Z, YIN G, et al Adaptive bilateral texture filter for image smoothing[J]. Frontiers in Neurorobotics, 2022, 16: 729924
doi: 10.3389/fnbot.2022.729924
1 XU L, YAN Q, XIA Y, et al Structure extraction from texture via relative total variation[J]. ACM Transactions on Graphics, 2012, 31 (6): 1- 10
2 CHEN Y, LI J, ZHOU Y Hyperspectral image denoising by total variation-regularized bilinear factorization[J]. Signal Processing, 2020, 174: 107645
doi: 10.1016/j.sigpro.2020.107645
30 ZHANG Q, SHEN X, XU L, et al. Rolling guidance filter [C]// European Conference on Computer Vision . [S.l.]: Springer, 2014: 815–830.
3 GHOSH S, CHAUDHURY K N On fast bilateral filtering using fourier kernels[J]. IEEE Signal Processing Letters, 2016, 23 (5): 570- 573
doi: 10.1109/LSP.2016.2539982
4 SONG C, XIAO C, LI X, et al Structure-preserving texture filtering for adaptive image smoothing[J]. Journal of Visual Languages and Computing, 2018, 45: 17- 23
doi: 10.1016/j.jvlc.2018.02.002
5 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): 1- 10
6 JIANG W, YANG X, WU W, et al Medical images fusion by using weighted least squares filter and sparse representation[J]. Computers and Electrical Engineering, 2018, 67: 252- 266
doi: 10.1016/j.compeleceng.2018.03.037
7 NAGAMANI K, DIVYA K, SUJATHA K, et al Adaptive histogram equalization of wavelet sub bands for the enhancement of contrast in aerial images[J]. Materials Today: Proceedings, 2018, 52: 898- 901
8 LIU Y P, ZHANG F, ZHANG Y, et al Image smoothing based on histogram equalized content-aware patches and direction-constrained sparse gradients[J]. Signal Processing, 2021, 183: 108037
doi: 10.1016/j.sigpro.2021.108037
9 YIN H, GONG Y, QIU G, et al Fast and efficient implementation of image filtering using a side window convolutional neural network[J]. Signal Processing, 2020, 176: 107717
doi: 10.1016/j.sigpro.2020.107717
10 GONG Y, TANG W, ZHOU L, et al. Quarter laplacian filter for edge aware image processing [C]// 2021 IEEE International Conference on Image Processing . Anchorage: IEEE, 2021: 1959–1963.
11 VDABO K, FOI A, KATKOVNIK V, et al Image denoising by sparse 3-D transform-domain collaborative filtering[J]. IEEE Transactions on Image Processing, 2007, 16 (8): 2080- 2095
doi: 10.1109/TIP.2007.901238
12 ZHANG X, HE C, Robust double-weighted guided image filtering [J]. Signal Processing , 2022, 199: 108609.
13 LIU W, ZHANG P, HUANG X, et al Real-time image smoothing via iterative least squares[J]. ACM Transactions on Graphics, 2020, 39 (3): 1- 24
14 SUN G, LIU S, WANG W, et al Dynamic range compression and detail enhancement algorithm for infrared image[J]. Applied Optics, 2014, 53 (26): 6013- 6029
doi: 10.1364/AO.53.006013
15 LI Z, ZHENG J Visual-salience-based tone mapping for high dynamic range images[J]. IEEE Transactions on Industrial Electronics, 2014, 61 (12): 7076- 7082
doi: 10.1109/TIE.2014.2314066
16 LEE W U, PARK S, KO S J. Image fusion-based tone mapping using gaussian mixture model clustering [C]// 2020 IEEE International Conference on Consumer Electronics . Las Vegas: IEEE, 2020: 1–4.
17 LI H, CHAN T N, QI X, et al Detail-preserving multi-exposure fusion with edge-preserving structural patch decomposition[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2021, 31 (11): 4293- 4304
doi: 10.1109/TCSVT.2021.3053405
18 LI Z, ZHENG J, RAHARDJA S Detail-enhanced exposure fusion[J]. IEEE Transactions on Image Processing, 2012, 21 (11): 4672- 4676
doi: 10.1109/TIP.2012.2207396
[1] 周雕,熊馨,周建华,宗静,张琪. 卷积神经网络结合子域适应的低采样率肌电手势识别[J]. 浙江大学学报(工学版), 2024, 58(10): 2011-2019.
[2] 江佳鸿,夏楠,李长吾,于鑫淼. 基于知识共享的遮挡人体姿态估计网络[J]. 浙江大学学报(工学版), 2024, 58(10): 2001-2010.
[3] 杜丁健,高遵海,陈倬. 基于改进YOLOv5的枸杞虫害检测[J]. 浙江大学学报(工学版), 2024, 58(10): 1992-2000.
[4] 李凡,杨杰,冯志成,陈智超,付云骁. 基于图像识别的弓网接触点检测方法[J]. 浙江大学学报(工学版), 2024, 58(9): 1801-1810.
[5] 冉庆东,郑力新. 基于改进YOLOv5的锂电池极片缺陷检测方法[J]. 浙江大学学报(工学版), 2024, 58(9): 1811-1821.
[6] 尹雅博,朱小飞,刘议丹. 源域数据增强与多兴趣细化迁移的跨域推荐模型[J]. 浙江大学学报(工学版), 2024, 58(8): 1717-1727.
[7] 李灿林,王新玥,马利庄,邵志文,张文娇. 融合注意力机制和结构线提取的图像卡通化[J]. 浙江大学学报(工学版), 2024, 58(8): 1728-1737.
[8] 李忠良,陈麒,石琳,杨朝,邹先明. 时间感知组合的动态知识图谱补全[J]. 浙江大学学报(工学版), 2024, 58(8): 1738-1747.
[9] 李林睿,王东升,范红杰. 基于法条知识的事理型类案检索方法[J]. 浙江大学学报(工学版), 2024, 58(7): 1357-1365.
[10] 杨军,张琛. 基于边界点估计与稀疏卷积神经网络的三维点云语义分割[J]. 浙江大学学报(工学版), 2024, 58(6): 1121-1132.
[11] 李运堂,李恒杰,张坤,王斌锐,关山越,陈源. 基于新型编码解码网络的复杂输电线识别[J]. 浙江大学学报(工学版), 2024, 58(6): 1133-1141.
[12] 邢志伟,朱书杰,李彪. 基于改进图卷积神经网络的航空行李特征感知[J]. 浙江大学学报(工学版), 2024, 58(5): 941-950.
[13] 魏翠婷,赵唯坚,孙博超,刘芸怡. 基于改进Mask R-CNN与双目视觉的智能配筋检测[J]. 浙江大学学报(工学版), 2024, 58(5): 1009-1019.
[14] 何勇禧,韩虎,孔博. 基于多依赖图和知识融合的方面级情感分析模型[J]. 浙江大学学报(工学版), 2024, 58(4): 737-747.
[15] 曹寅,秦俊平,高彤,马千里,任家琪. 基于生成对抗网络的文本两阶段生成高质量图像方法[J]. 浙江大学学报(工学版), 2024, 58(4): 674-683.