Lossy Image Compression with Compressive Autoencoders

TL;DR

This paper introduces a novel, efficient autoencoder-based method for lossy image compression that outperforms existing approaches like JPEG 2000 and is suitable for high-resolution images.

Contribution

It presents a simple modification to autoencoder training that achieves competitive compression performance with improved efficiency and scalability for high-resolution images.

Findings

Autoencoders can be effectively optimized for image compression with minimal loss function adjustments.

The proposed autoencoder outperforms JPEG 2000 and recent RNN-based methods.

The network is computationally efficient and suitable for high-resolution images.

Abstract

We propose a new approach to the problem of optimizing autoencoders for lossy image compression. New media formats, changing hardware technology, as well as diverse requirements and content types create a need for compression algorithms which are more flexible than existing codecs. Autoencoders have the potential to address this need, but are difficult to optimize directly due to the inherent non-differentiabilty of the compression loss. We here show that minimal changes to the loss are sufficient to train deep autoencoders competitive with JPEG 2000 and outperforming recently proposed approaches based on RNNs. Our network is furthermore computationally efficient thanks to a sub-pixel architecture, which makes it suitable for high-resolution images. This is in contrast to previous work on autoencoders for compression using coarser approximations, shallower architectures, computationally expensive methods, or focusing on small images.