结合天空分割和金字塔融合的多尺度图像去雾

doi:10.19678/j.issn.1000-3428.0065906

摘要/Abstract

摘要：

针对传统图像去雾方法存在对比度下降、色调偏暗和过度曝光的问题，提出一种结合天空分割和金字塔融合的多尺度图像去雾方法。利用含有梯度的分水岭算法分割天空区域，使用均值滤波计算天空区域得分，运用最高得分所对应最亮天空区域的平均灰度值获取更准确的大气光值。为了规避传统图像去雾方法估计的透射率强度值偏低或整体强度值偏高造成的负面影响，提出一种基于图像金字塔的多尺度透射率融合方法，通过拉普拉斯图像金字塔分别将暗通道先验和具有曝光特性的图像去雾算法估计的透射率图分解为多个尺度，采用感知融合方法融合拉普拉斯图像金字塔的每一层，重建融合透射率图的高斯图像金字塔，高斯图像金字塔的最底层即为融合后的透射率图；最后将所估计的大气光值和融合透射率代入大气散射模型，复原无雾图像。实验结果表明，所提去雾方法能够避免所复原无雾图像中存在的对比度下降、色调偏暗和过度曝光的问题，在SOTS室内外合成测试集上的PSNR和SSIM分别比排名第2的对比方法高出8.13%、0.59%和1.75%、1.57%。

关键词: 图像去雾, 天空分割, 透射率融合, 大气光值, 图像金字塔

Abstract:

A multiscale image dehazing method that combines sky segmentation and pyramid fusion is proposed to address contrast degradation, dark tones, and overexposure in traditional image dehazing methods. First, the method uses a watershed algorithm with gradients to segment the sky region, calculates the sky region score using mean filtering, and uses the average gray value of the brightest sky region corresponding to the highest score as the atmospheric light value. Second, to avoid the negative impacts caused by the low or high overall intensity values of the estimated transmissions by traditional image dehazing methods, a multiscale transmission fusion method based on image pyramid is proposed, using Laplacian image pyramid to decompose the transmission map estimated by the dark channel a priori and the image dehazing algorithm with exposure characteristics into multiple scales respectively, using perceptual fusion method to fuse each layer of the Laplacian image pyramid, and then reconstructing the Gaussian image pyramid of the fused transmission map, and the bottom layer of the Gaussian image pyramid is the fused transmission map. Finally, the estimated atmospheric light value and fused transmission are fed into the atmospheric scattering model to recover the dehazed image. Experiments show that the proposed dehazing method can avoid contrast degradation, dark tones, and overexposure in dehazed images. The PSNR and SSIM on the SOTS indoor and outdoor synthetic test sets are 8.13%, 0.59%, and 1.75%, 1.57% higher than those of the second-ranked comparison method.

Key words: image dehazing, sky segmentation, transmittance fusion, atmosphere light value, image pyramid

肖新杰, 李志伟, 张楠楠, 孙雨晴, 周武能. 结合天空分割和金字塔融合的多尺度图像去雾[J]. 计算机工程, 2023, 49(12): 205-213, 223.

Xinjie XIAO, Zhiwei LI, Nannan ZHANG, Yuqing SUN, Wuneng ZHOU. Multiscale Image Dehazing Method Combining Sky Segmentation and Pyramid Fusion[J]. Computer Engineering, 2023, 49(12): 205-213, 223.

http://www.ecice06.com/CN/Y2023/V49/I12/205

图/表 9

图1 结合天空区域分割和透射率的图像去雾方法原理

Fig.1 Principle of image dehazing method combining sky region segmentation and transmittance

图2 天空区域得分图计算流程

Fig.2 Calculation procedure of sky area score map

图3 基于图像金字塔的多尺度感知融合方法

Fig.3 Multi-scale perceptual fusion method based on image pyramid

图4 不同方法生成的透射率图及复原结果比较

Fig.4 Comparison of the transmittance maps and recovery results generated by different methods

图5 在SOTS数据集上的主观评价对比实验

Fig.5 Subjective evaluation comparison experiments on the SOTS dataset

图6 在HSTS数据集上真实世界图像上的主观评价对比实验

Fig.6 Subjective evaluation comparison experiments on real-world images on the HSTS dataset

图7 在HSTS数据集上合成图像上的主观评价对比实验

Fig.7 Subjective evaluation comparison experiments on synthetic images on the HSTS dataset

参考文献 29

1	吕建威, 钱锋, 韩昊男, 等. 结合天空分割和雾气浓度估计的图像去雾. 光学精密工程, 2022, 30(4): 464- 477. URL
	LÜ J W, QIAN F, HAN H N, et al. Single image dehazing with sky segmentation and haze density estimation. Optics and Precision Engineering, 2022, 30(4): 464- 477. URL
2	陈子妍, 龙道银, 王霄, 等. 基于逆通道与改进引导滤波的暗通道去雾算法. 计算机工程, 2021, 47(6): 245- 252. URL
	CHEN Z Y, LONG D Y, WANG X, et al. Dark channel defogging algorithm based on inverse channel and improved guided filtering. Computer Engineering, 2021, 47(6): 245- 252. URL
3	李力, 唐晓霏, 刘雯, 等. 航拍图像暗通道先验去雾算法速度改进. 吉林大学学报(理学版), 2021, 59(1): 77- 84. URL
	LI L, TANG X F, LIU W, et al. Speed improvement of dark channel prior fog removal algorithm for aerial images. Journal of Jilin University(Science Edition), 2021, 59(1): 77- 84. URL
4	TAN R T. Visibility in bad weather from a single image[C]//Proceedings of 2008 IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2008: 1-8.
5	ZHU Q S, MAI J M, SHAO L. A fast single image haze removal algorithm using color attenuation prior. IEEE Transactions on Image Processing, 2015, 24(11): 3522- 3533. doi: 10.1109/TIP.2015.2446191
6	BERMAN D, TREIBITZ T, AVIDAN S. Non-local image dehazing[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2016: 1674-1682.
7	CAI B L, XU X M, JIA K, et al. DehazeNet: an end-to-end system for single image haze removal. IEEE Transactions on Image Processing, 2016, 25(11): 5187- 5198. doi: 10.1109/TIP.2016.2598681
8	LI B Y, PENG X L, WANG Z Y, et al. AOD-Net: all-in-one dehazing network[C]//Proceedings of 2017 IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2017: 4770-4778.
9	QU Y Y, CHEN Y Z, HUANG J Y, et al. Enhanced Pix2Pix dehazing network[C]//Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2019: 8160-8168.
10	NARASIMHAN S G, NAYAR S K. Vision and the atmosphere. International Journal of Computer Vision, 2002, 48(3): 233- 254. doi: 10.1023/A:1016328200723
11	MCCARTNEY E J, HALL F. Optics of the atmosphere: scattering by molecules and particles. New York, USA: John Wiley & Sons Inc., 1976.
12	CHENG F C, CHENG C C, LIN P H, et al. A hierarchical airlight estimation method for image fog removal. Engineering Applications of Artificial Intelligence, 2015, 43, 27- 34. doi: 10.1016/j.engappai.2015.03.011
13	HE K M, SUN J, TANG X O. Single image haze removal using dark channel prior. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(12): 2341- 2353. doi: 10.1109/TPAMI.2010.168
14	HE K M, SUN J, TANG X O. Guided image filtering. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(6): 1397- 1409. doi: 10.1109/TPAMI.2012.213
15	MENG G F, WANG Y, DUAN J Y, et al. Efficient image dehazing with boundary constraint and contextual regularization[C]//Proceedings of 2013 IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2013: 617-624.
16	高银, 云利军, 石俊生, 等. 基于四阶PDE模型的暗原色理论雾天图像增强算法. 南京理工大学学报, 2015, 39(5): 544- 549. URL
	GAO Y, YUN L J, SHI J S, et al. Enhancement dark channel theory algorithm of fog image based on fourth-order PDE model. Journal of Nanjing University of Science and Technology, 2015, 39(5): 544- 549. URL
17	SHIN J, PARK H, PAIK J. Region-based dehazing via dual-supervised triple-convolutional network. IEEE Transactions on Multimedia, 2022, 24, 245- 260. doi: 10.1109/TMM.2021.3050053
18	QIN X, WANG Z L, BAI Y C, et al. FFA-net: feature fusion attention network for single image dehazing. Artificial Intelligence, 2020, 34(7): 11908- 11915.
19	JU M Y, DING C, REN W Q, et al. IDE: image dehazing and exposure using an enhanced atmospheric scattering model. IEEE Transactions on Image Processing, 2021, 30, 2180- 2192. doi: 10.1109/TIP.2021.3050643
20	BUCHSBAUM G. A spatial processor model for object colour perception. Journal of the Franklin Institute, 1980, 310(1): 1- 26. doi: 10.1016/0016-0032(80)90058-7
21	ZHAO S Y, ZHANG L, SHEN Y, et al. RefineDNet: a weakly supervised refinement framework for single image dehazing. IEEE Transactions on Image Processing, 2021, 30, 3391- 3404. doi: 10.1109/TIP.2021.3060873
22	GEUSEBROEK J M, VAN DEN BOOMGAARD R, SMEULDERS A W M, et al. Color and scale: the spatial structure of color images. Berlin, Germany: Springer, 2000.
23	GEUSEBROEK J M, VAN DEN BOOMGAARD R, SMEULDERS A W M, et al. Color invariance. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2001, 23(12): 1338- 1350. doi: 10.1109/34.977559
24	DUMOULIN V, VISIN F. A guide to convolution arithmetic for deep learning[EB/OL]. [2022-09-01]. https://arxiv.org/abs/1603.07285.pdf.
25	张楠楠, 李志伟, 郭新军, 等. 使用改进型大气散射模型的双阶段图像修复. 光学精密工程, 2022, 30(18): 2267- 2279. doi: 10.37188/OPE.20223018.2267
	ZHANG N N, LI Z W, GUO X J, et al. Two-stage image restoration using improved atmospheric scattering model. Optics and Precision Engineering, 2022, 30(18): 2267- 2279. doi: 10.37188/OPE.20223018.2267
26	王凯, 李志伟, 朱成德, 等. 基于二次引导滤波的局部立体匹配算法. 激光与光电子学进展, 2019, 56(8): 081004. URL
	WANG K, LI Z W, ZHU C D, et al. Local stereo matching algorithm based on secondary guided filtering. Laser & Optoelectronics Progress, 2019, 56(8): 081004. URL
27	LI B Y, REN W Q, FU D P, et al. Benchmarking single-image dehazing and beyond. IEEE Transactions on Image Processing, 2019, 28, 492- 505. doi: 10.1109/TIP.2018.2867951
28	WANG Z, BOVIK A C, SHEIKH H R, et al. Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing, 2004, 13, 600- 612. doi: 10.1109/TIP.2003.819861
29	宋传鸣, 唐媛, 乔明泽, 等. 缓解平滑区域光晕和偏色失真的图像去雾算法. 计算机辅助设计与图形学学报, 2022, 34(6): 953- 969. URL
	SONG C M, TANG Y, QIAO M Z, et al. Image dehazing algorithm for relieving halo effect and color distortion in smooth regions. Journal of Computer-Aided Design & Computer Graphics, 2022, 34(6): 953- 969. URL

[1]	卢昂, 储珺, 冷璐. 基于高低频特征增强的图像去雾[J]. 计算机工程, 2023, 49(8): 174-181.
[2]	王可铮, 徐玉芬, 周尚波. 结合对比感知损失和融合注意力的图像去雾模型[J]. 计算机工程, 2023, 49(8): 207-214.
[3]	朱钦权, 王同罕, 贾惠珍. 基于雾密度感知增强的去雾图像质量评价方法[J]. 计算机工程, 2023, 49(6): 208-216,226.
[4]	王效灵, 胡志杰, 徐帅帅, 黄浩如. 改进暗通道先验和策略性融合的图像去雾算法[J]. 计算机工程, 2023, 49(10): 212-221.
[5]	丁泳钧, 黄山. 一种用于图像去雾的改进生成对抗网络[J]. 计算机工程, 2022, 48(6): 207-212.
[6]	何涛, 俞舒曼, 徐鹤. 基于条件生成对抗网络与知识蒸馏的单幅图像去雾方法[J]. 计算机工程, 2022, 48(4): 165-172.
[7]	郭梦琰, 张娟, 刘巧红, 蔡立志. 基于循环生成对抗网络的图像去雾算法[J]. 计算机工程, 2022, 48(3): 280-287.
[8]	王国栋, 邵鹏, 王国宇, 刘少禹, 张建涛. 基于低秩分解与像素置乱的图像去雾方法[J]. 计算机工程, 2022, 48(12): 212-217.
[9]	黄山, 贾俊. 基于改进循环生成式对抗网络的图像去雾方法[J]. 计算机工程, 2022, 48(12): 218-223,231.
[10]	兰云伟, 崔智高, 苏延召, 汪波, 王念, 李艾华. 基于双尺度特征融合的单幅图像去雾网络[J]. 计算机工程, 2022, 48(11): 231-239.
[11]	温立民, 巨永锋, 王会峰, 徐娟. 基于B-样条曲线暗通道的图像去雾算法[J]. 计算机工程, 2022, 48(11): 215-223.
[12]	倪萍, 赖惠成, 高古学, 谷腾飞, 贾振红. 基于形态学与梯度域导向滤波的图像去雾算法[J]. 计算机工程, 2022, 48(10): 252-261.
[13]	赵慧, 魏伟波, 潘振宽, 纪连顺. 基于暗原色先验与变分正则化的图像去雾研究[J]. 计算机工程, 2021, 47(10): 214-220.
[14]	时璇, 冯舒婷, 沈传科, 李晨, 李党超. 透射率与场景亮度分块优化的图像去雾算法[J]. 计算机工程, 2019, 45(9): 235-241,247.
[15]	向文鼎, 杨平, 许冰. 基于自适应阈值分割与透射率融合的图像去雾算法[J]. 计算机工程, 2019, 45(10): 239-245.

选择文件类型/文献管理软件名称

选择包含的内容