Qonvolution: Towards Learning High-Frequency Signals with Queried Convolution

2025-12-14

AI Summary

Modeling high-frequency signals remains a fundamental challenge for neural networks in vision and graphics, hindered by spectral bias and optimization difficulties. This paper introduces *queried convolution* (Qonvolution), a query-driven lightweight convolutional mechanism that dynamically modulates local convolutional kernels using spatial queries—e.g., coordinates—without predefined basis functions or explicit positional encoding, thereby circumventing traditional spectral bias limitations. Qonvolution integrates seamlessly with Gaussian splatting and end-to-end differentiable optimization, making it naturally suited for 1D/2D regression, super-resolution, and novel view synthesis. In novel view synthesis, its integration with Gaussian splatting achieves state-of-the-art performance, significantly outperforming advanced radiance field methods. Across diverse tasks, Qonvolution consistently improves PSNR and SSIM, demonstrating superior high-frequency signal representation capability and strong generalization.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Non-convex Optimization
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Accurately learning high-frequency signals is a challenge in computer vision and graphics, as neural networks often struggle with these signals due to spectral bias or optimization difficulties. While current techniques like Fourier encodings have made great strides in improving performance, there remains scope for improvement when presented with high-frequency information. This paper introduces Queried-Convolutions (Qonvolutions), a simple yet powerful modification using the neighborhood properties of convolution. Qonvolution convolves a low-frequency signal with queries (such as coordinates) to enhance the learning of intricate high-frequency signals. We empirically demonstrate that Qonvolutions enhance performance across a variety of high-frequency learning tasks crucial to both the computer vision and graphics communities, including 1D regression, 2D super-resolution, 2D image regression, and novel view synthesis (NVS). In particular, by combining Gaussian splatting with Qonvolutions for NVS, we showcase state-of-the-art performance on real-world complex scenes, even outperforming powerful radiance field models on image quality.

Problem

• Enhances learning of high-frequency signals using queried convolutions
• Improves performance in vision and graphics tasks like super-resolution
• Achieves state-of-the-art results in novel view synthesis with Gaussian splatting

Innovation

• Queried-Convolutions enhance high-frequency signal learning
• Convolves low-frequency signals with queries like coordinates
• Combines with Gaussian splatting for novel view synthesis