一种基于信息保留网络的图像去噪算法

由于成像设备等各种因素影响，图像在成像或传感过程中会受到噪声干扰。图像去噪旨在减少或消除噪声对图像的影响，这一过程往往会导致高频信息的丢失。为了在去除图像噪声的同时保护图像的边缘信息与纹理细节，文章提出了一种计算复杂度相对较低的含有信息保留模块的卷积神经网络，直接对含噪声图像进行降噪。信息保留模块通过残差学习提取局部长路径和局部短路径的混合特征信息。该文采用峰值信噪比(PSNR/dB)和结构相似性(SSIM)两项评价指标对实验结果进行量化，这两项指标值越大，说明去噪效果越好。实验结果表明，在峰值信噪比和结构相似性2项评价指标的均值可达到30.36 dB和0.828 0，相比其他对比算法，2项评价指标分别平均提升了2.15 dB和0.072 9。该算法对不同种类、不同水平的噪声都具有良好的去噪效果，且速度优于所对比的一般算法，对基于卷积神经网络的去噪工作的进一步发展有一定的作用。

Abstract

Due to various factors such as imaging equipment, the image will be disturbed by noise during imaging or sensing. Image denoising aims to reduce or eliminate the influence of noise on the image, which often leads to the loss of high-frequency information. In order to protect the edge information and texture details of the image while removing image noise, a convolution neural network with information preservation blocks with relatively low computational complexity is proposed to denoise the noisy image directly. The information preservation block extracts the mixed feature information of the local long path and the local short path by residual learning. Peak signal to noise ratio (PSNR/dB) and structural similarity index method (SSIM) are used to quantify the experimental results. The larger the two indexes, the better the denoising effect. Experiments show that the mean values of PSNR and SSIM can reach 30.36 dB and 0.828 0. Compared with other denoising algorithms, the two evaluation indexes are improved by 2.15 dB and 0.072 9 respectively. The proposed algorithm has good denoising effect for different kinds and different levels of noise，and the speed is better than the general algorithms compared, which contributes to the further development of the denoising based on convolutional neural networks.

参考文献

[1] 吴海兵, 张良, 顾国华,等. 基于低照度三基色图像去噪及融合彩色图像增强方法研究［J］. 应用光学, 2018,39(1):57-63.

WU Haibing, ZHANG Liang, GU Guohua, et al. Color image enhancement based on LLL tricolor image denoising and fusion ［J］. Journal of Applied Optics , 2018，39 (1): 57-63.

[2] 王敏, 周磊, 周树道,等. 基于峰值信噪比和小波方向特性的图像奇异值去噪技术［J］. 应用光学, 2013, 34(1):85-89.

WANG Min, ZHOU Lei, ZHOU Shudao, et al. Image SVD denoising based on PSNR and wavelet directional feature ［J］. Journal of Applied Optics, 2013, 34 (1): 85-89.

[3] BIJALWAN A, GOYAL A, SETHI N. Wavelet transform based image denoise using threshold approaches［J］. International Journal of Engineering & Advanced Technology, 2012(5):218-221.

[4] AHARON M, ELAD M, BRUCKSTEIN A. The K-SVD: an algorithm for designing overcomplete dictionaries for sparse representation［J］. IEEE Transactions on Signal Processing, 2006, 54(11):4311-4322.

[5] DONG W, ZHANG L, SHI G, et al. Nonlocally centralized sparse representation for image restoration［J］. IEEE Transactions on Image Processing A Publication of the IEEE Signal Processing Society, 2013, 22(4):1620.

[6] 焦莉娟, 王文剑, 赵青杉,等. 近邻局部OMP稀疏表示图像去噪［J］. 中国图象图形学报, 2017, 22(11):1486-1492.

JIAO Lijuan, WANG Wenjian, ZHAO Qingshan, et al. Nearest neighbor local OMP sparse representation for image denoising ［J］. Journal of Image and Graphics, 2017, 22 (11): 1486-1492.

[7] DABOV 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.

[8] BANSAL M, DEVI M, JAIN N, et al. A proposed approach for biomedical image denoising using PCA_NLM［J］. International Journal of Bio-Science & Bio-Technology, 2014, 6(6):13-20.

[9] 韩震,王红斌, 余正涛,等.双边非局部均值滤波图像去噪算法［J］.传感器与微系统, 2016, 35(6):124-127.

HAN Zhen, WANG Hongbin, YU Zhengtao, et al. Bilateral non-local mean filter image denoising algorithm ［J］. Transducer and Microsystem Technologies, 2016, 35 (6): 124-127.

[10] HARMELING S, SCHULER C J, BURGER H C. Image denoising: Can plain neural networks compete with BM3D［C］//Providence,RI,USA：IEEE Conference on Computer Vision and Pattern Recognition. IEEE Computer Society, 2012:2392-2399.

[11] PAREKH A, SELESNICK I W. Convex denoising using non-convex tight frame regularization［J］. IEEE Signal Processing Letters, 2015, 22(10):1786-1790.

[12] ZORAN D, WEISS Y. From learning models of natural image patches to whole image restoration［J］. 2011, 6669(5):479-486.

[13] 高凯珺, 孙韶媛, 姚广顺,等. 基于深度学习的无人车夜视图像语义分割［J］. 应用光学, 2017, 38(3):421-428.

GAO Kaijun, SUN Shaoyuan, YAO Guangshun, et al. Semantic segmentation of night vision images based on depth learning ［J］. Journal of Applied Optics, 2017, 38 (3): 421-428.

[14] MARTIN D R, FOWLKES C C, MALIK J. Learning to detect natural image boundaries using local brightness, color, and texture cues［J］.IEEE/Transactions on Pattern Analysis and Machine Intelligence,2004,26(5):530-549.

[15] BEVILACQUA M, ROUMY A, GUILLEMOT C, et al. Single-image super-resolution via linear mapping of interpolated self-examples［J］. IEEE Transactions on Image Processing, 2014, 23(12):5334-47.

[16] ZHANG K, GAO X, TAO D, et al. Single image super-resolution with non-local means and steering kernel regression［J］. IEEE Transactions on Image Processing, 2012, 21(11):4544-4556.

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): 440. CHEN Qingjiang, SHI Xiaohan, CHAI Yuzhou. Image denoising algorithm based on information preservation network[J]. Journal of Applied Optics, 2019, 40(3): 440.

一种基于信息保留网络的图像去噪算法

关于本站 Cookie 的使用提示

全站搜索

一种基于信息保留网络的图像去噪算法

相关论文

相关资讯

关于本站 Cookie 的使用提示

全站搜索