Fast Sign Retrieval via Sub-band Convolution: An Elementary Extension of Binary Classification

2025-04-30

AI Summary

This paper addresses the problem of sign information loss when only magnitudes of block-wise DCT coefficients are retained. Recovering signs from magnitudes is a highly ill-posed, subband-structured binary classification task. To tackle it, we propose an efficient sign reconstruction method: first, we organize magnitude and sign values of same-frequency-band DCT coefficients into 3D subband blocks to explicitly capture frequency-domain locality and inter-band correlations; second, we design a lightweight 3D CNN architecture that jointly extracts spatial–spectral structural features within these blocks. Experiments demonstrate that our method achieves high reconstruction accuracy with minimal computational overhead—significantly outperforming conventional heuristic approaches and full-parameter models. By enabling accurate, low-cost sign recovery directly in the DCT domain, our work establishes a novel paradigm for DCT-domain sparse representation and compression.

Technology Category

• Data Mining & Knowledge Management: Data Compression
• Computer Vision: Representation Learning for Vision
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Security and Privacy: Applications of cryptography
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

To efficiently compress the sign information of images, we address a sign retrieval problem for the block-wise discrete cosine transformation~(DCT): reconstruction of the signs of DCT coefficients from their amplitudes. To this end, we propose a fast sign retrieval method on the basis of binary classification machine learning. We first introduce 3D representations of the amplitudes and signs, where we pack amplitudes/signs belonging to the same frequency band into a 2D slice, referred to as the sub-band block. We then retrieve the signs from the 3D amplitudes via binary classification, where each sign is regarded as a binary label. We implement a binary classification algorithm using convolutional neural networks, which are advantageous for efficiently extracting features in the 3D amplitudes. Experimental results demonstrate that our method achieves accurate sign retrieval with an overwhelmingly low computation cost.

Problem

• Reconstruct signs of DCT coefficients from amplitudes
• Propose fast sign retrieval via binary classification
• Use CNN for efficient 3D amplitude feature extraction

Innovation

• Uses sub-band blocks for 3D representation
• Employs binary classification for sign retrieval
• Implements convolutional neural networks efficiently