学习驱动的图像压缩算法研究

doi:10.19678/j.issn.1000-3428.0068782

摘要/Abstract

Abstract: 目前，基于CNN（卷积神经网络）深度学习的图像压缩已经取得了非常优异的成果，但是由于其感受野受限，不能学习到图像非局部区域内像素之间的上下文关系，缺少了长距离建模和感知能力，造成结果出现失真、伪影和压缩率较高等问题。为了解决以上问题，本文提出了以下两种解决办法：第一，本文设计了一种由CNN、多尺度注意力机制 (Multi scale attention，MSA)和残差单元所构成的对称编解码器架构，该架构在对图片进行通道变换和空间变换的同时引入多尺度注意力机制，能够对特征进行重新校准，减少潜在表示的冗余像素。第二：设计了一种基于U型框架的超先验网络，可以在不同层级上获取多尺度的上下文信息，在帮助提取高级语义特征的同时，保留详细的低级特征信息，能够更好的进行边界细化和细节恢复。在Kodak，Tecnick和CLIC三种数据集上，与其他先进的图像压缩方法进行对比实验，在相同比特率下，本文提出的模型在PSNR指标上分别提高了约0.3 dB、0.6 dB、0.5 dB。并且在保证压缩率的同时，有效的提高了对非重复纹理特征和图像细节特征的重建效果。

杨红菊, 吉昌. 学习驱动的图像压缩算法研究[J]. 计算机工程, doi: 10.19678/j.issn.1000-3428.0068782.

YANG HongJu, JI Chang. learning-driven image compression algorithm[J]. Computer Engineering, doi: 10.19678/j.issn.1000-3428.0068782.

参考文献

[1] Wallace G K. The JPEG still picture compression standard[J]. Communications of the ACM, 1991, 34(4): 30-44.
[2] Taubman D S, Marcellin M W, Rabbani M. JPEG2000: Image compression fundamentals, standards and practice[J]. Journal of Electronic Imaging, 2002, 11(2): 286-287.
[3] Fabrice Bellard. Bpg image format, 2014. https:// bellard.org/bpg/. 1
[4] 叶宗苗. 基于深度学习的端到端智能图像压缩研究 [D].杭州电子科技大学,2022.DOI:10.27075/d.cnki.ghz dc.202 2.000096. (Ye Zongmia. End-to-End Intelligent Image Compressio n Based on Deep Learning[D]. Hangzhou Dianzi Unive rsi, 2022.DOI:10.27075/d.cnki.ghz dc.202 2.000096.
[5] Lin F, Sun H, Liu J, et al. Multistage spatial context models for learned image compression[C]//ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2023: 1-5.
[6] Wang D, Yang W, Hu Y, et al. Neural data-dependent transform for learned image compression[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022: 17379-17388.
[7] Fu H, Liang F, Lin J, et al. Learned image compression with discretized gaussian-laplacian-logistic mixture model and concatenated residual modules[J]. arXiv preprint arXiv:2107.06463, 2021.
[8] Cheng Z, Sun H, Takeuchi M, et al. Learned image compression with discretized gaussian mixture likelihood s and attention modules[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 7939-7948.
[9] Ballé J, Laparra V, Simoncelli E P. End-to-end optimized image compression[J]. arXiv preprint arXiv:1611.01704, 2016.
[10] Ballé J, Laparra V, Simoncelli E P. Density modeling of images using a generalized normalization transformation [J]. arXiv preprint arXiv:1511.06281, 2015.
[11] Minnen D, Singh S. Channel-wise autoregressive entropy models for learned image compression[C]//2020 IEEE International Conference on Image Processing (ICIP). IEEE, 2020: 3339-3343.
[12] Guo Z, Zhang Z, Feng R, et al. Causal contextual prediction for learned image compression[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2021, 32(4): 2329-2341.
[13] He D, Yang Z, Peng W, et al. Elic: Efficient learned image compression with unevenly grouped space-channel contextual adaptive coding[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022: 5718-5727.
[14] Mentzer F, Toderici G D, Tschannen M, et al. High-fidelity generative image compression[J]. Advances in Neural Information Processing Systems, 2020, 33: 11913-11924.
[15] Xie Y, Cheng K L, Chen Q. Enhanced invertible encoding for learned image compression[C]//Proceedings of the 29th ACM international conference on multimedia. 2021: 162-170.
[16] 皇甫晓瑛,钱惠敏,黄敏.结合注意力机制的深度神经网络综述[J].计算机与现代化,2023(02):40-49+57. (HUANGFU Xiao-ying，QIAN Hui-min，HUANG Min. A Review of Deep Neural Networks Combined with Attention Mechanism[J]. JISUANJI YU XIANDAIHUA, 2023(02):40-49+57.)
[17] Zhao L, Bai H, Wang A, et al. Learning a virtual codecbased on deep convolutional neural network to compress image[J]. Journal of Visual Communication and Image Representation, 2019, 63: 102589.
[18] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[J]. Advances in neural information processing systems, 2017, 30.
[19] Carion N, Massa F, Synnaeve G, et al. End-to-end object detection with transformers[C]//European conference on computer vision. Cham: Springer International Publishin g, 2020: 213-229.
[20] Zou R, Song C, Zhang Z. The devil is in the details: Window-based attention for image compression[C]// Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2022: 17492-17501.
[21] Qian Y, Lin M, Sun X, et al. Entroformer: A transformer-based entropy model for learned image compression[J]. arXiv preprint arXiv:2202.05492, 2022.
[22] Kim J H, Heo B, Lee J S. Joint global and local hierarchical priors for learned image compression[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022: 5992-6001.
[23] Ho J, Jain A, Abbeel P. Denoising diffusion probabilistic models[J]. Advances in neural information processing systems, 2020, 33: 6840-6851.
[24] He D, Zheng Y, Sun B, et al. Checkerboard context model for efficient learned image compression[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021: 14771-14780.
[25] Kingma D P, Ba J. Adam: A method for stochastic optimization[J]. arXiv preprint arXiv:1412.6980, 2014.
[26] Kodak E. Kodak lossless true color image suite (PhotoCD PCD0992) [J]. URL http://r0k. us/graphics/kodak, 1993, 6.
[27] Asuni N, Giachetti A. TESTIMAGES: a Large-scale Archive for Testing Visual Devices and Basic Image Processing Algorithms[C]//STAG. 2014: 63-70.
[28] Toderici G, Shi W, Timofte R, et al. Workshop and challenge on learned image compression [C]//Proceed ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020.
[29] Ballé J, Minnen D, Singh S, et al. Variational image compression with a scale hyperprior[J]. arXiv preprint arXiv:1802.01436, 2018.
[30] Chen F, Xu Y, Wang L. Two-stage octave residual network for end-to-end image compression[C]// Proceed ings of the AAAI Conference on Artificial Intelligence. 2022, 36(4): 3922-3929.
[31] He D, Yang Z, Peng W, et al. Elic: Efficient learned image compression with unevenly grouped space-channel contextual adaptive coding[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022: 5718-5727.
[32] Minnen D, Ballé J, Toderici G D. Joint autoregressive and hierarchical priors for learned image compression[J]. Advances in neural information processing systems, 2018, 31.
[33] Joint Video Experts Team. Vvc official test model vtm. 2021.
[34] Luo W, Li Y, Urtasun R, et al. Understanding the effective receptive field in deep convolutional neural networks[J]. Advances in neural information processing systems, 2016, 29.
[35] Liu J, Sun H, Katto J. Learned image compression with mixed transformer-cnn architectures[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2023: 14388-14397.

选择文件类型/文献管理软件名称

选择包含的内容