Multi-Frame Blind Manifold Deconvolution for Rotating Synthetic Aperture Imaging

2025-01-31

AI Summary

Blind deconvolution in Rotational Synthetic Aperture (RSA) imaging is severely challenged by motion blur arising from multi-angle rotational motion, leading to ill-posed image restoration. Method: This paper proposes a manifold-prior-driven multi-frame joint blind deconvolution framework. It introduces the low-dimensional manifold structure of high-dimensional latent images into the RSA blind deconvolution formulation—jointly leveraging manifold fitting and manifold regularization to encode geometric priors of the latent scene, thereby departing from conventional Maximum A Posteriori (MAP)-based probabilistic modeling. The method integrates nonlinear manifold learning, multi-frame blind convolutional modeling, and efficient iterative optimization. Contribution/Results: It significantly improves pixel-wise intensity estimation accuracy and structural detail preservation. Simulation results demonstrate superior sharpness, enhanced texture fidelity, and overall performance exceeding state-of-the-art deconvolution algorithms.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Learning with Manifolds
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Security and Privacy: Large-scale security measurements

Abstract

Rotating synthetic aperture (RSA) imaging system captures images of the target scene at different rotation angles by rotating a rectangular aperture. Deblurring acquired RSA images plays a critical role in reconstructing a latent sharp image underlying the scene. In the past decade, the emergence of blind convolution technology has revolutionised this field by its ability to model complex features from acquired images. Most of the existing methods attempt to solve the above ill-posed inverse problem through maximising a posterior. Despite this progress, researchers have paid limited attention to exploring low-dimensional manifold structures of the latent image within a high-dimensional ambient-space. Here, we propose a novel method to process RSA images using manifold fitting and penalisation in the content of multi-frame blind convolution. We develop fast algorithms for implementing the proposed procedure. Simulation studies demonstrate that manifold-based deconvolution can outperform conventional deconvolution algorithms in the sense that it can generate a sharper estimate of the latent image in terms of estimating pixel intensities and preserving structural details.

Problem

• Multi-image
• Rotational Deblurring
• Synthetic Aperture Imaging

Innovation

• Simple Structure Matching
• Penalty Scoring Method
• Blind Deconvolution