On Self-Adaptive Perception Loss Function for Sequential Lossy Compression

2025-02-15

AI Summary

This work addresses the low perceptual fidelity, error accumulation, and insufficient exploitation of temporal correlations in low-latency lossy video compression. To this end, we propose PLF-SA—a causal and adaptive perceptual loss function. PLF-SA jointly models the distribution of the current source frame and previously reconstructed frames, dynamically adapting reconstruction quality to suppress error propagation and enhance temporal consistency under distortion constraints. Theoretically, we establish, for the first time, a rate-distortion-perception bound incorporating temporal dependencies, rigorously derived under a first-order Markov source and Gaussian signal model. Experimentally, on Moving MNIST and UVG datasets, PLF-SA significantly improves visual realism and fine-detail fidelity over conventional perceptual losses while maintaining identical MSE distortion.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Machine Learning: Causal Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

We consider causal, low-latency, sequential lossy compression, with mean squared-error (MSE) as the distortion loss, and a perception loss function (PLF) to enhance the realism of reconstructions. As the main contribution, we propose and analyze a new PLF that considers the joint distribution between the current source frame and the previous reconstructions. We establish the theoretical rate-distortion-perception function for first-order Markov sources and analyze the Gaussian model in detail. From a qualitative perspective, the proposed metric can simultaneously avoid the error-permanence phenomenon and also better exploit the temporal correlation between high-quality reconstructions. The proposed metric is referred to as self-adaptive perception loss function (PLF-SA), as its behavior adapts to the quality of reconstructed frames. We provide a detailed comparison of the proposed perception loss function with previous approaches through both information theoretic analysis as well as experiments involving moving MNIST and UVG datasets.

Problem

• Enhancing realism in lossy compression
• Adapting to reconstructed frame quality
• Avoiding error-permanence in reconstructions

Innovation

• Self-adaptive perception loss function
• Joint distribution consideration
• Temporal correlation exploitation