Every Packet Counts: Dispersing Information for Loss-Resilient Learned Image Compression

2026-08-11

AI Summary

This work addresses the significant degradation in reconstruction quality of existing learned image compression methods under high packet loss scenarios, such as satellite and emergency communications. To enhance robustness, the authors propose an end-to-end packet-loss-resilient compression framework that redistributes critical information across channels via an Inter-Channel Redistribution (ICR) mechanism and disperses it through Interleaved Channel Grouping (ICG). Additionally, a dual-branch autoregressive entropy model is introduced to shorten decoding dependency chains. Notably, the model requires only uniform random packet loss during training yet achieves strong generalization to bursty loss patterns. Experimental results demonstrate that, at a 20% packet loss rate, the method improves average PSNR by 1.84 dB and reduces PSNR variance by an order of magnitude, outperforming state-of-the-art approaches specifically designed for bursty packet loss.

Abstract

Learned image compression (LIC) has achieved impressive rate-distortion performance. However, existing methods remain highly vulnerable to packet loss, a common challenge in satellite and emergency communications. This vulnerability stems from non-uniform information distribution at the packetization stage and sequential decoding dependencies at the entropy coding stage. We propose an end-to-end loss-resilient image compression scheme that addresses both. Before packetization, we introduce an Inter-Channel Redistribution (ICR) mechanism to redistribute channel energy, preventing critical information concentrating in a small subset of channels. Then, an Interleaved Channel Grouping (ICG) strategy partitions latent channels in a strided manner to disperse information across packets, with each packet kept within constrained sizes. To limit cascading errors from lost packets, we adopt a two-layer dual-branch autoregressive structure to shorten the dependency chain. Extensive experiments demonstrate that our method consistently outperforms existing approaches in both reconstruction quality and stability. At 20% packet loss, it achieves an average PSNR gain of 1.84 dB over LossResilientLIC while reducing PSNR variance by an order of magnitude. Notably, trained under uniform random loss only, our model generalizes to bursty loss modeled by the Gilbert-Elliott channel, outperforming methods explicitly trained for such conditions.

Problem

• packet loss
• learned image compression
• loss resilience
• information distribution
• decoding dependency

Innovation

• Inter-Channel Redistribution
• Interleaved Channel Grouping
• loss-resilient compression
• autoregressive entropy coding
• learned image compression