Most learning-based image compression methods lack efficiency for high image quality due to their non-invertible design. The decoding function of the frequently applied compressive autoencoder architecture is only an approximated inverse of the encoding transform. This issue can be resolved by using invertible latent variable models, which allow a perfect reconstruction if no quantization is performed. Furthermore, many traditional image and video coders apply dynamic block partitioning to vary the compression of certain image regions depending on their content. Inspired by this approach, hierarchical latent spaces have been applied to learning-based compression networks. In this paper, we present a novel concept, which adapts the hierarchical latent space for augmented normalizing flows, an invertible latent variable model. Our best performing model achieved average rate savings of more than 7% over comparable single-scale models.

翻译：大多数基于学习的图像压缩方法由于非可逆的设计，在高图像质量方面效率不足。常用的压缩自编码器架构的解码函数仅是编码变换的近似逆。这一问题可通过使用可逆潜变量模型来解决，该模型在不进行量化的情况下允许完美重建。此外，许多传统图像和视频编码器根据图像区域内容，采用动态块分割来调整其压缩程度。受此方法启发，层级潜空间已被应用于基于学习的压缩网络。本文提出了一种新概念，将层级潜空间适配于增强正规化流——一种可逆潜变量模型。我们的最佳模型相较于可比单尺度模型平均节省了超过7%的码率。