VibrantLeaves: A principled parametric image generator for training deep restoration models

2025-04-14

AI Summary

Deep neural networks exhibit poor generalization and limited robustness in image restoration, primarily due to implicit biases toward natural training data and insufficient modeling of underlying physical mechanisms. To address this, we propose a parameterized synthetic image generator grounded in the Dead Leaves framework. Our method systematically decouples and quantifies the independent contributions of geometric structure, texture statistics, and imaging physics to restoration performance—marking the first such systematic analysis. The generator enables interpretable and controllable synthesis via explicit geometric modeling, parametric texture generation, and simplified optical imaging simulation. Experiments on denoising and super-resolution demonstrate that models trained solely on our synthetic data achieve performance comparable to those trained on natural images (PSNR/SSIM degradation <0.3 dB). Moreover, they exhibit over 20% improved robustness against geometric (e.g., rotation, scaling) and radiometric (e.g., brightness variation) perturbations.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Even though Deep Neural Networks are extremely powerful for image restoration tasks, they have several limitations. They are poorly understood and suffer from strong biases inherited from the training sets. One way to address these shortcomings is to have a better control over the training sets, in particular by using synthetic sets. In this paper, we propose a synthetic image generator relying on a few simple principles. In particular, we focus on geometric modeling, textures, and a simple modeling of image acquisition. These properties, integrated in a classical Dead Leaves model, enable the creation of efficient training sets. Standard image denoising and super-resolution networks can be trained on such datasets, reaching performance almost on par with training on natural image datasets. As a first step towards explainability, we provide a careful analysis of the considered principles, identifying which image properties are necessary to obtain good performances. Besides, such training also yields better robustness to various geometric and radiometric perturbations of the test sets.

Problem

• Addressing biases in deep neural networks for image restoration
• Creating synthetic training sets for better model control
• Enhancing robustness to geometric and radiometric perturbations

Innovation

• Parametric image generator for synthetic training sets
• Geometric modeling and texture integration
• Simple image acquisition modeling for robustness