首页 > 论文 > 激光与光电子学进展 > 55卷 > 3期(pp:31012--1)

基于多特征融合的卷积神经网络图像去雾算法

Convolution Neural Network Image Defogging Based on Multi-Feature Fusion

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

针对传统去雾算法需要人工提取特征,对比度低、信噪比低等问题, 提出一种基于多特征融合的卷积神经网络去雾算法。利用卷积神经网络算法模拟人类视觉系统对雾天图像进行层次化处理, 实现自动提取特征。算法采用直接从雾天图像到清晰无雾图像映射的学习方式, 该映射由特征提取、多尺度特征融合和浅层深层特征融合联合实现。多尺度特征融合提升网络对图像细节的重建, 浅层深层特征融合则将浅层卷积得到的轮廓信息和深层卷积得到的细节信息进行融合, 提升去雾重建的整体效果。实验结果表明, 相比于单一尺度网络, 多特征融合网络的峰值信噪比提高了1.280 dB。本文算法对自然雾天图像去雾效果明显, 细节信息和对比度均优于其他算法, 为去雾方法的研究提供了新思路。

Abstract

We propose a convolutional neural network defogging algorithm based on multi-feature fusion to overcome the problem of manual feature extraction, low contrast, and low signal-to-noise ratio in traditional defogging algorithms. The convolution neural network simulates the human visual system to hierarchically process the fog images and automatically extract image features. The algorithm adopts a learning method of the direct mapping from the hazing image to the clear defogging image, which includes feature extraction, multi-scale feature fusion, and shallow and deep feature fusion. Multi-scale feature fusion helps to rebuild details of the image. Shallow and deep feature fusion combines the contour information obtained by shallow convolution with the detail information obtained by deep convolution to enhance the overall effect of defogging. The experimental results show that the peak signal to noise ratio of the multi-feature fusion network increases by 1.280 dB compared with the single-scale network. The proposed algorithm has obvious defogging effect on natural fog image and superior detail information and contrast compared with other algorithms, which provides a new idea for defogging methods.

广告组1 - 空间光调制器+DMD
补充资料

中图分类号:TP751.1

DOI:10.3788/LOP55.031012

所属栏目:图像处理

基金项目:国家自然科学基金(61372145)、天津大学自主创新基金(2015XZC-0005)

收稿日期:2017-09-26

修改稿日期:2017-10-16

网络出版日期:--

作者单位    点击查看

徐岩:天津大学电气自动化与信息工程学院, 天津 300072
孙美双:天津大学电气自动化与信息工程学院, 天津 300072

联系人作者:孙美双(18722621468@163.com)

备注:孙美双(1991-), 女, 硕士研究生, 主要从事图像处理、模式识别、深度学习等方面的研究。E-mail: 18722621468@163.com

【1】Lu S H, Huang H Z, Zhang H Z, et al. Comprehensive performance evaluation for video surveillance cameras[J]. Laser & Optoelectronics Progress, 2015, 52(9): 091102.
卢树华, 黄鸿志, 张鸿洲, 等. 视频监控摄像机性能参数综合评价研究[J]. 激光与光电子学进展, 2015, 52(9): 091102.

【2】Zhang S H, Chen H F, Zhang H T, et al. Design of optical lens for traffic surveillance system[J]. Laser & Optoelectronics Progress, 2011, 48(2): 022202.
张善华, 陈慧芳, 张海艇, 等. 用于交通监控系统的光学成像镜头的设计[J]. 激光与光电子学进展, 2011, 48(2): 022202.

【3】Li S J, Fan X, Zhu B, et al. A method for small infrared targets detection based on the technology of motion blur recovery[J]. Acta Optica Sinica, 2017, 37(6): 0610001.
李思俭, 樊祥, 朱斌, 等. 基于运动模糊复原技术的红外弱小目标检测方法[J]. 光学学报, 2017, 37(6): 0610001.

【4】Wu Y Q, Yin D Y. Detection of small target in infrared image based on background predication by FLS-SVM[J]. Acta Optica Sinica, 2010, 30(10): 2806-2811.
吴一全, 尹丹艳. 基于FLS-SVM背景预测的红外弱小目标检测[J]. 光学学报, 2010, 30(10): 2806-2811.

【5】Stark J A. Adaptive image contrast enhancement using generalizations of histogram equalization[J]. IEEE Transactions on Image Processing, 2000, 9(5): 889-896.

【6】Xu Z Y, Liu X M, Ji N. Fog removal from color images using contrast limited adaptive histogram equalization[C]// 2nd International Congress on Image and Signal Processing, 2009: 10955452.

【7】Yang W T, Wang R G, Fan S, et al. Variable filter Retinex algorithm for foggy image enhancement[J]. Journal of Computer-Aided Design & Computer Graphics, 2010, 22(6): 965-971.
杨万挺, 汪荣贵, 方帅, 等. 滤波器可变的Retinex雾天图像增强算法[J]. 计算机辅助设计与图形学学报, 2010, 22(6): 965-971.

【8】Yu P, Hao C C. Fractional differential and multiscale Retinex combined foggy image enhancement[J]. Laser & Optoelectronics Progress, 2018, 55(1): 011012.
余萍, 郝成成. 分数阶微分和多尺度Retinex联合的雾霭图像增强[J]. 激光与光电子学进展, 2018, 55(1): 011012.

【9】Tan R T. Visibility in bad weather from a single image[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2008: 1-8.

【10】Fattal R. Single image dehazing[J]. ACM Transactions on Graphics, 2008, 27(3): 72.

【11】He K M, Sun J, Tang X O. Single image haze removal using dark channel prior[C]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(12): 2341.

【12】Tarel J P, Hautière N. Fast visibility restoration from a single color or gray level image[C]. IEEE 12th International Conference on Computer Vision, 2009: 2201-2208.

【13】Sun X M, Sun J X, Zhao L R, et al. Improved algorithm for single image haze removing using dark channel prior[J]. Journal of Image and Graphics, 2014, 19(3): 381-385.
孙小明, 孙俊喜, 赵立荣, 等. 暗原色先验单幅图像去雾改进算法[J]. 中国图象图形学报, 2014, 19(3): 381-385.

【14】Yang A P, Bai H H. Nighttime image defogging based on the theory of Retinex and dark channel prior[J]. Laser & Optoelectronics Progress, 2017, 54(4): 041002.
杨爱萍, 白煌煌. 基于Retinex理论和暗通道先验的夜间图像去雾算法[J]. 激光与光电子学进展, 2017, 54(4): 041002.

【15】He K M, Zhang X Y, Ren S Q, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(9): 1904-1916.

【16】Li T, Sun J G, Zhang X J, et al. Research of hyperspectral image classification based on hierarchical sparse representation feature learning[J]. Laser & Optoelectronics Progress, 2016, 53(9): 091001.
李铁, 孙劲光, 张新君, 等. 基于分层稀疏表示特征学习的高光谱图像分类研究[J]. 激光与光电子学进展, 2016, 53(9): 091001.

【17】Ouyang W L, Luo P, Zeng X Y, et al. DeepID-Net: multi-stage and deformable deep convolutional neural networks for object detection[J]. Eprint Arxiv, 2014.

【18】Ye G L, Sun S Y, Gao K J, et al. Nighttime pedestrian detection based on faster region convolution neural network[J]. Laser & Optoelectronics Progress, 2017, 54(8): 081003.
叶国林, 孙韶媛, 高凯珺, 等. 基于加速区域卷积神经网络的夜间行人检测研究[J]. 激光与光电子学进展, 2017, 54(8): 081003.

【19】Liu S, Liang X D, Liu L Q, et al. Matching-CNN meets KNN: quasi-parametric human parsing[J]. IEEE Conference on Computer Vision and Pattern Recognition, 2015: 1419-1427.

【20】Cai B L, Xu X M, Jia K, et al. DehazeNet: an end-to-end system for single image haze removal[J]. IEEE Transactions on Image Processing, 2016, 25(11): 5187-5198.

【21】McCartney E J. Optics of the atmosphere: scattering by molecules and particles[J]. IEEE Journal of Quantum Electronics, 1977, 24(7): 779-779.

【22】Ma M, Li Y B. Multi-level image sequences and convolutional neural networks based human action recognition method[J]. Journal of Jilin University Engineering and Technology Edition, 2017, 47(4): 1244-1252.
马淼, 李贻斌. 基于多级图像序列和卷积神经网络的人体行为识别[J]. 吉林大学学报(工学版), 2017, 47(4): 1244-1252.

【23】Durbin R, Rumelhart D E. Product units: a computationally powerful and biologically plausible extension to backpropagation networks[J]. Neural Computation, 1989, 1(1): 133-142.

【24】Liu W J, Liang X J, Qu H C. Learning performance of convolutional neural networks with different pooling models[J]. Journal of Image and Graphics, 2016, 21(9): 1178-1190.
刘万军, 梁雪剑, 曲海成. 不同池化模型的卷积神经网络学习性能研究[J]. 中国图象图形学报, 2016, 21(9): 1178-1190.

【25】Zeiler M D, Fergus R. Visualizing and understanding convolutional networks[J]. Springer International Publishing, 2014, 8689: 818-833.

【26】Lecun Y, Bottou L, Bengio Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.

【27】Scharstein D, Szeliski R, Zabih R. A taxonomy and evaluation of dense two-frame stereo correspondence algorithms[J]. International Journal of Computer Vision, 2002, 47(1/2/3): 7-42.

【28】Scharstein D, Szeliski R. High-accuracy stereo depth maps using structured light[C]. IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003: 195-202.

【29】Scharstein D, Pal C. Learning conditional random fields for stereo[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2007: 9738080.

【30】Hirschmuller H, Scharstein D. Evaluation of cost functions for stereo matching[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2007: 9738137.

【31】Dong C, Loy C C, He K M, et al. Image super-resolution using deep convolutional networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(2): 295-307.

【32】Kwon H H, Tai Y W, Lin S. Data-driven depth map refinement via multi-scale sparse representation[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2015: 159-167.

【33】Meng G, Wang Y, Duan J, et al. Efficient image dehazing with boundary constraint and contextual regularization[C]. IEEE International Conference on Computer Vision, 2013: 617-624.

【34】Galdran A, Vazquez-Corral J, Pardo D, et al. Fusion-based variational image dehazing[J]. IEEE Signal Processing Letters, 2017, 24(2): 151-155.

【35】Huynh-Thu Q, Barkowsky M, Callet P L. The importance of visual attention in improving the 3D-TV viewing experience: overview and new perspectives[J]. IEEE Transactions on Broadcasting, 2011, 57(2): 421-431.

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF