Low-Rank Adaptation of Pre-Trained Stable Diffusion for Rigid-Body Target ISAR Imaging

2025-03-26

AI Summary

To address the imaging blurring in rigid-body ISAR caused by low time-frequency resolution and strong noise in the Range-Instantaneous-Doppler (RID) method, this paper proposes a time-frequency representation super-resolution enhancement technique based on LoRA-finetuned Stable Diffusion Turbo, integrated into the RID imaging pipeline. It is the first work to apply a low-rank adaptation (LoRA)-based generative diffusion model to radar time-frequency spectrogram super-resolution and joint denoising. The approach enables training on simulated data while achieving cross-domain generalization to real-world measurements. Through adversarial training and joint optimization of RID and ISAR imaging, it significantly improves Doppler frequency estimation accuracy and image focus: time-frequency resolution is enhanced by over 2×, and noise suppression reaches 18.7 dB.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Traditional range-instantaneous Doppler (RID) methods for rigid-body target imaging often suffer from low resolution due to the limitations of time-frequency analysis (TFA). To address this challenge, our primary focus is on obtaining high resolution time-frequency representations (TFRs) from their low resolution counterparts. Recognizing that the curve features of TFRs are a specific type of texture feature, we argue that pre trained generative models such as Stable Diffusion (SD) are well suited for enhancing TFRs, thanks to their powerful capability in capturing texture representations. Building on this insight, we propose a novel inverse synthetic aperture radar (ISAR) imaging method for rigid-body targets, leveraging the low-rank adaptation (LoRA) of a pre-trained SD model. Our approach adopts the basic structure and pre-trained parameters of SD Turbo while incorporating additional linear operations for LoRA and adversarial training to achieve super-resolution and noise suppression. Then we integrate LoRA-SD into the RID-based ISAR imaging, enabling sharply focused and denoised imaging with super-resolution capabilities. We evaluate our method using both simulated and real radar data. The experimental results demonstrate the superiority of our approach in frequency es timation and ISAR imaging compared to traditional methods. Notably, the generalization capability is verified by training on simulated radar data and testing on measured radar data.

Problem

• Enhance low-resolution time-frequency representations for rigid-body ISAR imaging
• Adapt pre-trained Stable Diffusion model for super-resolution and noise suppression
• Integrate LoRA-SD into RID-based ISAR imaging for sharper, denoised results

Innovation

• LoRA-adapted Stable Diffusion for TFR enhancement
• Adversarial training for super-resolution and denoising
• Integration of LoRA-SD into RID-based ISAR imaging