Lossy image compression methods with deep neural network (DNN) include a quantization process between encoder and decoder networks as an essential part to increase the compression rate. However, the quantization operation impedes the flow of gradient and often disturbs the optimal learning of the encoder, which results in distortion in the reconstructed images. To alleviate this problem, this paper presents a simple yet effective way that enhances the performance of lossy image compression without imposing training overhead or modifying the original network architectures. In the proposed method, we utilize an auxiliary branch called a shortcut which directly connects the encoder and decoder. Since the shortcut does not include the quantization process, it supports the optimal learning of the encoder by flowing the accurate gradient. Furthermore, to assist the decoder which should handle additional feature maps obtained via the shortcut, we also propose a residual refinement unit (RRU) following the quantizer. The experimental results show that the image compression network trained with the proposed method remarkably improves the performance in terms of peak signal-to-noise ratio (PSNR), structural similarity (SSIM), and multi-scale structural similarity (MS-SSIM).
Bibliographical noteFunding Information:
Manuscript received January 16, 2020; revised March 17, 2020; accepted March 18, 2020. Date of publication March 23, 2020; date of current version April 30, 2020. This work was supported by the Institute for Information & communications Technology Promotion (IITP) grant funded by the Korea government (MSIT) (2017-0-00250, Intelligent Defense Boundary Surveillance Technology Using Collaborative Reinforced Learning of Embedded Edge Camera and Image Analysis). The associate editor coordinating the review of this manuscript and approving it for publication was Prof. Mylene Q. Farias. (Corresponding author: Sung-Jea Ko.) The authors are with the School of Electrical Engineering Department, Korea University, Sungbuk-gu, Seoul 136-713, South Korea (e-mail: yjyeo@dali. korea.ac.kr; email@example.com; firstname.lastname@example.org; swkim@ dali.korea.ac.kr; email@example.com). Digital Object Identifier 10.1109/LSP.2020.2982561
© 1994-2012 IEEE.
- Convolutional neural network
- Deep learning
- Image compression
ASJC Scopus subject areas
- Signal Processing
- Electrical and Electronic Engineering
- Applied Mathematics