一种基于双分支改良编解码器的图像去噪算法

亓法国;张海洋;柳淳;赵长明;张子龙

应用光学 ›› 2020, Vol. 41 ›› Issue (5) : 956-964.

应用光学 ›› 2020, Vol. 41 ›› Issue (5) : 956-964. DOI: 10.5768/JAO202041.0502004

一种基于双分支改良编解码器的图像去噪算法

  • 亓法国1, 张海洋1, 柳淳1, 赵长明1, 张子龙1
作者信息 +

Image denoising algorithm based on dual-branch modified codec

  • QI Faguo1, ZHANG Haiyang1, LIU Chun1, ZHAO Changming1, ZHANG Zilong1
Author information +
文章历史 +

摘要

针对传统图像去噪算法多噪声去除难,深层卷积神经网络去噪模型网络复杂、训练时间长等问题,提出一种基于自编码器结构的双分支改良编解码网络,实现高效图像去噪。双分支结构之一采用降-升采样实现点噪声消除,另一分支专注于宏观的图像修复和伪像去除,后端利用残差结构进行整合,实现数字图像混合噪声去噪。实验结果显示:对于含有标准差为15,均值为0的高斯噪声、噪声密度为5%的椒盐噪声和散粒噪声的混合噪声图像测试集,实验去噪效果相较于输入混合噪声图像峰值信噪比,平均提升了5.3%。与12层全卷积神经网络相比,去噪效果相当,训练速度提升了25.4%,体现了其“轻量级”的优点。实验表明:该方法相较于深层卷积神经网络,训练速度快,网络简单;相较于传统图像去噪算法,噪声去除效果也较为明显。该算法可应用于轻量级视觉平台后端去噪。

Abstract

Aiming at the problems of the traditional image denoising algorithm such as difficult multi-noise removal, complex deep convolutional neural network denoising model network and long training time, a dual-branch modified codec(DMC) network based on auto-encoder structure was proposed to achieve the high-efficient image denoising. One of the dual branch structure used the down-up sampling to eliminate the point noise, the other focused on the macroscopical image restoration and artifacts removal, and the residual structure was used to integrate at the end to realize the mixed noise denoising of the digital image. The experimental results show that for the image test set of the mixed noise containing Gaussian noise with standard deviation of 15 and mean value of 0, salt and pepper noise as well as shot noise with noise density of 5%, compared with the peak signal-to-noise ratio of the input mixed noise image, the experimental denoising effect is improved by 5.3% on average. Compared with the 12-layer full convolutional neural network, the denoising effect is equivalent and the training speed is increased by about 25.4%, which embodies the advantages of its lightweight. The experimental conclusions indicate that compared with the deep convolution neural network, this method has the advantages of fast training speed and simple network; compared with the traditional image denoising algorithm, it has better noise removal effect. This algorithm can be applied to the end denoising of lightweight vision platform.

关键词

双分支编解码 / 轻量级 / 残差 / 图像去噪

Key words

image denoising / residual / dual branch codec / lightweight

引用本文

导出引用
亓法国, 张海洋, 柳淳, 赵长明, 张子龙. 一种基于双分支改良编解码器的图像去噪算法. 应用光学. 2020, 41(5): 956-964 https://doi.org/10.5768/JAO202041.0502004
QI Faguo, ZHANG Haiyang, LIU Chun, ZHAO Changming, ZHANG Zilong. Image denoising algorithm based on dual-branch modified codec. Journal of Applied Optics. 2020, 41(5): 956-964 https://doi.org/10.5768/JAO202041.0502004

基金

国防科工局重大基础科研项目(JCKY2016201A601)

参考文献

李传朋, 秦品乐, 张晋京. 基于深度卷积神经网络的图像去噪研究[J]. 计算机工程,2017,43(03):253-260.
张嘉超. 低照度数字相机处理流水线噪声抑制方法研究[D]. 南京: 南京理工大学, 2018.
李轩. CMOS图像传感器噪声抑制研究[D]. 天津: 天津大学, 2010.
蔡红苹. 基于小波变换的图像去噪方法研究[D]. 长沙: 国防科学技术大学, 2003.
陈清江, 石小涵, 柴昱洲. 一种基于信息保留网络的图像去噪算法[J]. 应用光学,2019,40(3):88-94.
孔峻. 深度卷积神经网络在计算机视觉中的应用[J]. 电子技术与软件工程,2018,143(21):146-147.
周飞燕, 金林鹏, 董军. 卷积神经网络研究综述[J]. 计算机学报,2017,40(6):1229-1251.
孙娅楠. 梯度下降法在机器学习中的应用[D]. 成都: 西南交通大学, 2018.
陈振宏, 兰艳艳, 郭嘉丰, 等. 基于差异合并的分布式随机梯度下降算法[J]. 计算机学报,2015,38(10):2054-2063.

LI Chuanpeng, QIN Pinle, ZHANG Jinjing. Research on image denoising based on deep convolutional neural network[J]. Computer Engineering,2017,43(03):253-260.
ZHANG B, KAGAWA K, TAKASAWA T, et al. Low-light color reproduction by selective averaging in multi-aperture camera with bayer color-filter low-noise CMOS image sensors[J]. ITE Transactions on Media Technology and Applications,2015,3(4):234-239.
BOUKHAYMA A, PEIZERAT A, ENZ C. Temporal readout noise analysis and reduction techniques for low-light CMOS image sensors[J]. IEEE Transactions on Electron Devices,2016,63(1):72-78.
ZHANG Jiachao. Research on noise suppression method of low-light digital camera processing pipeline[D]. Nanjing: Nanjing University of Science and Technology, 2018.
LI Xuan. Research on CMOS image sensor noise suppression[D]. Tianjin: Tianjin University, 2010.
CAI Hongping. Research on image denoising method based on wavelet transform[D]. Changsha: National University of Defense Technology, 2003.
JAIN V, SEUNG H

Accesses

Citation

Detail

段落导航
相关文章

/