Learned Image Compression with Soft Bit-based Rate-Distortion Optimization
This work improves image compression for applications requiring high-quality storage or transmission, but it is incremental as it builds on existing learning-based methods.
The paper tackled the problem of rate-distortion optimization in learning-based image compression by introducing soft bits to address zero gradients from quantization and inaccurate rate estimation, achieving state-of-the-art performance in MS-SSIM and PSNR metrics.
This paper introduces the notion of soft bits to address the rate-distortion optimization for learning-based image compression. Recent methods for such compression train an autoencoder end-to-end with an objective to strike a balance between distortion and rate. They are faced with the zero gradient issue due to quantization and the difficulty of estimating the rate accurately. Inspired by soft quantization, we represent quantization indices of feature maps with differentiable soft bits. This allows us to couple tightly the rate estimation with context-adaptive binary arithmetic coding. It also provides a differentiable distortion objective function. Experimental results show that our approach achieves the state-of-the-art compression performance among the learning-based schemes in terms of MS-SSIM and PSNR.