GAN-based Image Compression with Improved RDO Process

Fanxin Xia,Jian Jin,Lili Meng,Feng Ding,Huaxiang Zhang

2023-06-18

Abstract:GAN-based image compression schemes have shown remarkable progress lately due to their high perceptual quality at low bit rates. However, there are two main issues, including 1) the reconstructed image perceptual degeneration in color, texture, and structure as well as 2) the inaccurate entropy model. In this paper, we present a novel GAN-based image compression approach with improved rate-distortion optimization (RDO) process. To achieve this, we utilize the DISTS and MS-SSIM metrics to measure perceptual degeneration in color, texture, and structure. Besides, we absorb the discretized gaussian-laplacian-logistic mixture model (GLLMM) for entropy modeling to improve the accuracy in estimating the probability distributions of the latent representation. During the evaluation process, instead of evaluating the perceptual quality of the reconstructed image via IQA metrics, we directly conduct the Mean Opinion Score (MOS) experiment among different codecs, which fully reflects the actual perceptual results of humans. Experimental results demonstrate that the proposed method outperforms the existing GAN-based methods and the state-of-the-art hybrid codec (i.e., VVC).

Image and Video Processing,Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

This paper attempts to solve two main problems existing in the image compression schemes based on Generative Adversarial Networks (GANs):

1. **Perceptual degradation of reconstructed images**: The perceptual quality declines in terms of color, texture, and structure. Although the existing GAN - based methods perform well at low bit rates, they still have deficiencies in these aspects.
2. **Inaccurate entropy models**: The entropy models in the existing methods cannot accurately estimate the probability distribution of the latent representation, thus affecting the compression performance.

To solve these problems, the author proposes an improved GAN - based image compression method, which specifically includes the following:

- **Introducing DISTS and MS - SSIM metrics**: They are used to evaluate the perceptual degradation of reconstructed images in terms of color, texture, and structure. These two metrics can more comprehensively reflect the quality changes of images.
- **Adopting the discretized Gaussian - Laplace - Logistic Mixture Model (GLLMM)**: It is used for entropy modeling to improve the accuracy of probability distribution estimation and reduce the reconstruction error caused by inaccurate entropy estimation.

Through these improvements, the author hopes to significantly improve the perceptual quality of reconstructed images while maintaining a low bit rate, and has proven in experiments that this method is superior to the existing GAN - based image compression methods and the state - of - the - art hybrid codecs (such as VVC).