9 January 2024 in IEEE Sensors Journal
Event-based structured light 3D measurement systems (eSL) overcome intrinsic limitations of frame-based structured light 3D measurement systems (fSL) such as sensor dynamic range, pixel bandwidth, and system optical power. However, most of the existing methods are based on laser scanning, and they are not only limited by the speed of acquisition, but are also very sensitive to noise events. This paper proposes an efficient event-driven fringe projection structured light 3D reconstruction system (eFPSL), aiming for high-precision 3D measurements at increased acquisition speeds. Leveraging a digital light processing fringe projection system, eFPSL enhances the acquisition speed. We further develop a time-frequency analysis algorithm capable of efficiently converting event streams into 2D fringe intensity images with high signal-to-noise ratio and excellent sinusoidality thus mitigating noise event impacts. Moreover, a novel events-count-based shadow mask extraction (eSME) strategy is proposed to eliminate shadow noise. Experimental validations demonstrate eFPSL’s ability to produce higher precision 3D reconstructions, reducing mean absolute error (MAE) by 64% compared to existing eSL methods. In addition, eFPSL impressively safeguards comprehensive 3D results while reducing system power consumption, achieving an 82% reduction in system optical power compared to traditional fSL systems. Most importantly, for high dynamic range (HDR) scenes, eFPSL achieves visual representation and measurement accuracy comparable to eight measurements of fSL systems using only a single measurement, indicating an 87.5% improvement in measurement efficiency. Hence, our experiments confirm eFPSL’s adaptability can extend to HDR scenes involving high-speed motion.
5 December 2022 in Optics and Lasers in Engineering
21 August 2025 in Physics in Medicine & Biology
Objective . Reconstructing high-quality images from corrupted measurements remains a fundamental challenge in medical imaging. Recently, deep unrolling (DUN) methods have emerged as a promising solution, combining the interpretability of traditional iterative algorithms with the powerful representation capabilities of deep learning. However, their performance is often limited by weak information flow between iterative stages and a constrained ability to capture global features across stages-limitations that tend to worsen as the number of iterations increases. Approach . In this work, we propose a memory-enhanced (ME) and multi-domain learning-based DUN network for interpretable, high-fidelity medical image reconstruction. First, a ME module is designed to adaptively integrate historical outputs across stages, reducing information loss. Second, we introduce a cross-stage spatial-domain learning transformer (CS-SLFormer) to extract both local and non-local features within and across stages, improving reconstruction performance. Finally, a frequency-domain consistency learning module ensures alignment between reconstructed and ground truth images in the frequency domain, recovering fine image details. Main results . Comprehensive experiments evaluated on three representative medical imaging modalities (positron emission tomography (PET), magnetic resonance imaging, and computed tomography) show that the proposed method consistently outperforms state-of-the-art approaches in both quantitative metrics and visual quality. Specifically, our method achieved a peak signal-to-noise of 37.835 dB and an structural similarity index of 0.970 in 1 dose PET reconstruction. Significance . This study expands the use of model-driven deep learning in medical imaging, demonstrating the potential of ME DUN frameworks for high-quality reconstructions.