Learning Better Lossless Compression Using Lossy Compression
This work addresses the problem of efficient lossless image compression for applications requiring high fidelity, though it is incremental as it builds on existing lossy compression techniques.
The paper tackles lossless image compression by using a lossy compression algorithm (BPG) to decompose images into a lossy reconstruction and a residual, then models the residual with a convolutional neural network for entropy coding, achieving state-of-the-art performance by outperforming previous learned methods and standards like PNG, WebP, and JPEG2000.
We leverage the powerful lossy image compression algorithm BPG to build a lossless image compression system. Specifically, the original image is first decomposed into the lossy reconstruction obtained after compressing it with BPG and the corresponding residual. We then model the distribution of the residual with a convolutional neural network-based probabilistic model that is conditioned on the BPG reconstruction, and combine it with entropy coding to losslessly encode the residual. Finally, the image is stored using the concatenation of the bitstreams produced by BPG and the learned residual coder. The resulting compression system achieves state-of-the-art performance in learned lossless full-resolution image compression, outperforming previous learned approaches as well as PNG, WebP, and JPEG2000.