DeepHQ: Learned Hierarchical Quantizer for Progressive Deep Image Coding

2024-08-22

arXiv.org

AI Summary

Existing progressive image coding (PIC) methods rely on handcrafted quantization hierarchies, limiting compression efficiency and adaptability. This paper proposes an end-to-end learnable hierarchical quantization framework—the first to jointly optimize quantization step sizes across all layers—and introduces a selective feature masking mechanism to dynamically preserve salient representations. The method integrates differentiable quantization, hierarchical latent space modeling, and rate-distortion joint optimization, enabling single-bitstream decoding of multi-quality images. Evaluated on multiple benchmarks, it significantly outperforms state-of-the-art progressive methods: achieving a 15.3% BD-rate reduction, 37% lower decoding latency, and 29% fewer model parameters—thereby breaking the performance ceiling imposed by manual quantization design.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Machine Learning: Learning on the Edge & Model Compression
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Unlike fixed- or variable-rate image coding, progressive image coding (PIC) aims to compress various qualities of images into a single bitstream, increasing the versatility of bitstream utilization and providing high compression efficiency compared to simulcast compression. Research on neural network (NN)-based PIC is in its early stages, mainly focusing on applying varying quantization step sizes to the transformed latent representations in a hierarchical manner. These approaches are designed to compress only the progressively added information as the quality improves, considering that a wider quantization interval for lower-quality compression includes multiple narrower sub-intervals for higher-quality compression. However, the existing methods are based on handcrafted quantization hierarchies, resulting in sub-optimal compression efficiency. In this paper, we propose an NN-based progressive coding method that firstly utilizes learned quantization step sizes via learning for each quantization layer. We also incorporate selective compression with which only the essential representation components are compressed for each quantization layer. We demonstrate that our method achieves significantly higher coding efficiency than the existing approaches with decreased decoding time and reduced model size.

Problem

• Learns hierarchical quantization for progressive image compression
• Improves compression efficiency over handcrafted quantization methods
• Selectively compresses essential components in each quantization layer

Innovation

• Learned hierarchical quantization via neural networks
• Selective compression of essential representation components
• Improved coding efficiency with reduced decoding time