Learned Image Compression with Separate Hyperprior Decoders
This work addresses an incremental improvement in image compression efficiency for applications requiring high-quality visual data storage and transmission.
The paper tackled the performance bottleneck in learned image compression caused by a single hyperprior decoder leading to a collapsed Gaussian model, and proposed using three separate hyperprior decoders to improve parameter estimation, resulting in an average 3.36% BD-rate reduction compared to state-of-the-art methods.
Learned image compression techniques have achieved considerable development in recent years. In this paper, we find that the performance bottleneck lies in the use of a single hyperprior decoder, in which case the ternary Gaussian model collapses to a binary one. To solve this, we propose to use three hyperprior decoders to separate the decoding process of the mixed parameters in discrete Gaussian mixture likelihoods, achieving more accurate parameters estimation. Experimental results demonstrate the proposed method optimized by MS-SSIM achieves on average 3.36% BD-rate reduction compared with state-of-the-art approach. The contribution of the proposed method to the coding time and FLOPs is negligible.