Optics and Lasers in Engineering最新文献_第6页

Machine-learning-enabled loop microcavity for multipoint sensing of microscale nonlinear thermal fields 用于微尺度非线性热场多点传感的基于机器学习的环形微腔

IF 3.7 2区工程技术 Q2 OPTICS

Optics and Lasers in Engineering

Pub Date : 2026-01-16 DOI: 10.1016/j.optlaseng.2026.109620

Yingdong He , Wei Liu , Jiahe Ouyang , Jianhui Zhong , Chengbin Li , Yi Li , Yun Lin , Hao Dai , Zhijun Wu , Xining Zhang

A PDMS-encapsulated microfiber loop cavity (MLC) temperature sensor combined with a random forest (RF) model is proposed to achieve precise multipoint temperature prediction within millimeter-scale micro-regions. By constructing anisotropic thermal fields using orthogonal heating wires, the MLC’s optical responses were analyzed to infer temperatures at multiple discrete locations, including on- and off-microfiber positions. The RF model, trained with structural parameters and integrated optical intensity, achieved high prediction accuracy (RMSE≈2.5°C, R²≈0.97 for the horizontal heating) across multiple sensing points. Temperature gradients and their vector characteristics were subsequently derived from the predicted temperatures, revealing distinct spatial characteristics under horizontal and vertical heating that are strongly correlated with device geometry. This study demonstrates that integrating optical microcavity sensing with machine learning enables stable thermal analysis without requiring multi-sensor arrays, offering a promising route for microelectronic thermal management, structural health monitoring, and high-temperature warning in micro-nano devices.

提出了一种结合随机森林（RF）模型的pdms封装的微光纤环路腔（MLC）温度传感器，可实现毫米尺度微区域内的多点精确温度预测。通过使用正交加热丝构建各向异性热场，分析MLC的光响应，以推断多个离散位置的温度，包括微光纤打开和关闭位置。利用结构参数和集成光强训练的射频模型在多个感测点上实现了较高的预测精度（RMSE≈2.5°C，水平加热时R2≈0.97）。温度梯度及其矢量特征随后从预测温度中得到，揭示了水平和垂直加热下的明显空间特征，这些特征与器件几何形状密切相关。该研究表明，将光学微腔传感与机器学习相结合，无需多传感器阵列即可实现稳定的热分析，为微纳器件的微电子热管理、结构健康监测和高温预警提供了一条有前途的途径。

{"title":"Machine-learning-enabled loop microcavity for multipoint sensing of microscale nonlinear thermal fields","authors":"Yingdong He , Wei Liu , Jiahe Ouyang , Jianhui Zhong , Chengbin Li , Yi Li , Yun Lin , Hao Dai , Zhijun Wu , Xining Zhang","doi":"10.1016/j.optlaseng.2026.109620","DOIUrl":"10.1016/j.optlaseng.2026.109620","url":null,"abstract":"<div><div>A PDMS-encapsulated microfiber loop cavity (MLC) temperature sensor combined with a random forest (RF) model is proposed to achieve precise multipoint temperature prediction within millimeter-scale micro-regions. By constructing anisotropic thermal fields using orthogonal heating wires, the MLC’s optical responses were analyzed to infer temperatures at multiple discrete locations, including on- and off-microfiber positions. The RF model, trained with structural parameters and integrated optical intensity, achieved high prediction accuracy (RMSE≈2.5°C, R<sup>2</sup>≈0.97 for the horizontal heating) across multiple sensing points. Temperature gradients and their vector characteristics were subsequently derived from the predicted temperatures, revealing distinct spatial characteristics under horizontal and vertical heating that are strongly correlated with device geometry. This study demonstrates that integrating optical microcavity sensing with machine learning enables stable thermal analysis without requiring multi-sensor arrays, offering a promising route for microelectronic thermal management, structural health monitoring, and high-temperature warning in micro-nano devices.</div></div>","PeriodicalId":49719,"journal":{"name":"Optics and Lasers in Engineering","volume":"201 ","pages":"Article 109620"},"PeriodicalIF":3.7,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145969393","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

DI-FP: Digital incoherent Fourier ptychography for far-field imaging 用于远场成像的数字非相干傅立叶平面摄影

IF 3.7 2区工程技术 Q2 OPTICS

Optics and Lasers in Engineering

Pub Date : 2026-01-15 DOI: 10.1016/j.optlaseng.2026.109601

Jinghao Xu , Yizheng Liao , Tianci Feng , Siyuan Wang , Duan Luo , An Pan

This paper proposes a novel Digital Incoherent Fourier Ptychography (DI-FP) technique that effectively addresses the speckle noise challenge in long-range Fourier ptychographic imaging through an innovative batch gradient summation mechanism. Compared with conventional methods, this study makes several key contributions: First, we develop a feature-domain batch gradient summation algorithm that exploits the randomness of multi-angle speckles to achieve automatic noise cancellation without requiring additional preprocessing. Second, we construct a new reconstruction framework integrating incoherent imaging with feature extraction, which significantly enhances image contrast while maintaining resolution. Experimental results demonstrate that for imaging at distances of 12.8m and 65m, our method improves reconstruction quality (PSNR) from 5.42dB (conventional method) to 13.98dB, substantially reduces speckle contrast, and decreases single reconstruction time from 150s to 44s. This work provides a new solution for long-range high-resolution optical imaging that combines excellent anti-noise performance with computational efficiency, showing significant application potential in remote sensing monitoring and target recognition fields.

本文提出了一种新的数字非相干傅立叶平面成像（DI-FP）技术，该技术通过一种创新的批梯度求和机制，有效地解决了长程傅立叶平面成像中的散斑噪声问题。与传统方法相比，本研究做出了以下几个关键贡献：首先，我们开发了一种特征域批量梯度求和算法，该算法利用多角度散斑的随机性，在不需要额外预处理的情况下实现自动降噪。其次，构建了非相干成像与特征提取相结合的重构框架，在保持分辨率的同时显著增强了图像对比度；实验结果表明，对于距离为12.8m和65m的成像，该方法将重建质量（PSNR）从传统方法的5.42dB提高到13.98dB，显著降低了散斑对比度，并将单次重建时间从150秒缩短到44秒。该研究为远程高分辨率光学成像提供了一种新的解决方案，该解决方案结合了优异的抗噪声性能和计算效率，在遥感监测和目标识别领域具有重要的应用潜力。

{"title":"DI-FP: Digital incoherent Fourier ptychography for far-field imaging","authors":"Jinghao Xu , Yizheng Liao , Tianci Feng , Siyuan Wang , Duan Luo , An Pan","doi":"10.1016/j.optlaseng.2026.109601","DOIUrl":"10.1016/j.optlaseng.2026.109601","url":null,"abstract":"<div><div>This paper proposes a novel Digital Incoherent Fourier Ptychography (DI-FP) technique that effectively addresses the speckle noise challenge in long-range Fourier ptychographic imaging through an innovative batch gradient summation mechanism. Compared with conventional methods, this study makes several key contributions: First, we develop a feature-domain batch gradient summation algorithm that exploits the randomness of multi-angle speckles to achieve automatic noise cancellation without requiring additional preprocessing. Second, we construct a new reconstruction framework integrating incoherent imaging with feature extraction, which significantly enhances image contrast while maintaining resolution. Experimental results demonstrate that for imaging at distances of 12.8m and 65m, our method improves reconstruction quality (PSNR) from 5.42dB (conventional method) to 13.98dB, substantially reduces speckle contrast, and decreases single reconstruction time from 150s to 44s. This work provides a new solution for long-range high-resolution optical imaging that combines excellent anti-noise performance with computational efficiency, showing significant application potential in remote sensing monitoring and target recognition fields.</div></div>","PeriodicalId":49719,"journal":{"name":"Optics and Lasers in Engineering","volume":"200 ","pages":"Article 109601"},"PeriodicalIF":3.7,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979641","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Weighted Multi-Interval spatial-temporal binary encoding fourier single-pixel imaging 加权多间隔时空二值编码傅立叶单像素成像

IF 3.7 2区工程技术 Q2 OPTICS

Optics and Lasers in Engineering

Pub Date : 2026-01-15 DOI: 10.1016/j.optlaseng.2025.109597

Jiangping Zhu , Yi Shuai , Pei Zhou , Guigang Yin , Lei Liu , Fangji Gan , Yanxia Zhou

Fourier Single-Pixel Imaging (FSI) captures an object’s frequency spectrum by modulating it with Fourier basis patterns. However, traditional grayscale patterns are challenging to project at high speeds using spatial light modulators. Existing binarization methods, which rely on upsampling and dithering, trade off spatial resolution for approximation. In this work, we develop a spatial-temporal binarization technique for FSI and evaluate its performance. Our method divides the intensity range into more intervals than the conventional K+1, and incorporates error diffusion along with weighted strategies. By preserving the sine-wave characteristics of the binary patterns, this approach significantly increases the information content, improving both encoding accuracy and reconstruction quality. Simulations and experiments demonstrate superior performance compared to existing methods with the same number of patterns: simulations show over a 10% improvement in SSIM and PSNR, with more than a 30% reduction in RMSE. Experimental results show up to an 11% increase in SSIM, a 7% improvement in PSNR, and a 13% reduction in RMSE, enabling more precise detail reconstruction.

傅立叶单像素成像（FSI）通过用傅立叶基模式调制物体的频谱来捕获物体的频谱。然而，传统的灰度模式是具有挑战性的投影在高速使用空间光调制器。现有的二值化方法依赖于上采样和抖动，牺牲了空间分辨率来实现近似。在这项工作中，我们开发了一种用于FSI的时空二值化技术并评估了其性能。该方法将强度范围划分为比传统K+1更多的区间，并结合误差扩散和加权策略。该方法通过保留二进制图形的正弦波特征，显著增加了信息含量，提高了编码精度和重建质量。仿真和实验表明，与具有相同模式数量的现有方法相比，该方法具有优越的性能：仿真显示，SSIM和PSNR提高了10%以上，RMSE降低了30%以上。实验结果表明，SSIM提高了11%，PSNR提高了7%，RMSE降低了13%，从而实现了更精确的细节重建。

{"title":"Weighted Multi-Interval spatial-temporal binary encoding fourier single-pixel imaging","authors":"Jiangping Zhu , Yi Shuai , Pei Zhou , Guigang Yin , Lei Liu , Fangji Gan , Yanxia Zhou","doi":"10.1016/j.optlaseng.2025.109597","DOIUrl":"10.1016/j.optlaseng.2025.109597","url":null,"abstract":"<div><div>Fourier Single-Pixel Imaging (FSI) captures an object’s frequency spectrum by modulating it with Fourier basis patterns. However, traditional grayscale patterns are challenging to project at high speeds using spatial light modulators. Existing binarization methods, which rely on upsampling and dithering, trade off spatial resolution for approximation. In this work, we develop a spatial-temporal binarization technique for FSI and evaluate its performance. Our method divides the intensity range into more intervals than the conventional K+1, and incorporates error diffusion along with weighted strategies. By preserving the sine-wave characteristics of the binary patterns, this approach significantly increases the information content, improving both encoding accuracy and reconstruction quality. Simulations and experiments demonstrate superior performance compared to existing methods with the same number of patterns: simulations show over a 10% improvement in SSIM and PSNR, with more than a 30% reduction in RMSE. Experimental results show up to an 11% increase in SSIM, a 7% improvement in PSNR, and a 13% reduction in RMSE, enabling more precise detail reconstruction.</div></div>","PeriodicalId":49719,"journal":{"name":"Optics and Lasers in Engineering","volume":"200 ","pages":"Article 109597"},"PeriodicalIF":3.7,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979644","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Terahertz rapid high-resolution imaging via inverse-designed line-focusing metalens 通过反设计线聚焦超透镜的太赫兹快速高分辨率成像

IF 3.7 2区工程技术 Q2 OPTICS

Optics and Lasers in Engineering

Pub Date : 2026-01-14 DOI: 10.1016/j.optlaseng.2025.109586

Zhuo Sun, Hao Hu, Xiaoxue Hu, Jiayu Chen, Xiaolei Wang

The industrial adoption of terahertz (THz) nondestructive testing is constrained by the inherent trade-off between imaging speed and spatial resolution. Conventional systems improve resolution by increasing the number of scanning points, but this inevitably results in exponentially longer acquisition times. Line-detector-based architectures shorten data collection, yet their performance remains limited when relying on traditional metasurfaces design approaches, which only allow local tuning of pre-established structures. Here, we propose an efficient inverse-design framework that integrates metasurfaces with an Adaptive Hybrid Optimization Algorithm (AHOA) to achieve rapid, high-resolution THz imaging. The framework dynamically adjusts the exploration range through mutation operators and leverages Particle Swarm Optimization to directly determine the structural parameters of the metasurfaces across the full design space. To validate this approach, we designed and fabricated a line-focusing metalens (LFM) operating at 0.1 THz for linear array detection. Experimental results show a focal spot with a full width at half maximum (FWHM) of 4.1 mm, along with a sidelobe energy of 42%, an energy efficiency of 49%, and a consistent 2 mm resolution maintained over a 20 mm depth of field. When integrated into a line-scanning system, the LFM enables single-pass acquisition of complete images, supporting imaging at 66.7 mm/s while preserving sub-diffraction-limit resolution. This design strategy is expected to provide a reference solution for inverse design in terahertz imaging.

太赫兹（THz）无损检测的工业应用受到成像速度和空间分辨率之间固有权衡的限制。传统的系统通过增加扫描点的数量来提高分辨率，但这不可避免地会导致更长的采集时间。基于线检测器的架构缩短了数据收集时间，但当依赖于传统的元表面设计方法时，它们的性能仍然有限，传统的元表面设计方法只允许预先建立的结构进行局部调优。在这里，我们提出了一个有效的反设计框架，将元表面与自适应混合优化算法（AHOA）集成在一起，以实现快速，高分辨率的太赫兹成像。该框架通过变异算子动态调整探测范围，并利用粒子群优化直接确定整个设计空间的超表面结构参数。为了验证这种方法，我们设计并制造了一个工作在0.1太赫兹的线聚焦超透镜（LFM），用于线阵列检测。实验结果表明，在20 mm的景深范围内，焦斑的全宽为4.1 mm，副瓣能量为42%，能量效率为49%，分辨率为2 mm。当集成到线扫描系统中时，LFM可以单次采集完整的图像，支持66.7 mm/s的成像，同时保持亚衍射极限分辨率。该设计策略有望为太赫兹成像的逆设计提供参考解决方案。

{"title":"Terahertz rapid high-resolution imaging via inverse-designed line-focusing metalens","authors":"Zhuo Sun, Hao Hu, Xiaoxue Hu, Jiayu Chen, Xiaolei Wang","doi":"10.1016/j.optlaseng.2025.109586","DOIUrl":"10.1016/j.optlaseng.2025.109586","url":null,"abstract":"<div><div>The industrial adoption of terahertz (THz) nondestructive testing is constrained by the inherent trade-off between imaging speed and spatial resolution. Conventional systems improve resolution by increasing the number of scanning points, but this inevitably results in exponentially longer acquisition times. Line-detector-based architectures shorten data collection, yet their performance remains limited when relying on traditional metasurfaces design approaches, which only allow local tuning of pre-established structures. Here, we propose an efficient inverse-design framework that integrates metasurfaces with an Adaptive Hybrid Optimization Algorithm (AHOA) to achieve rapid, high-resolution THz imaging. The framework dynamically adjusts the exploration range through mutation operators and leverages Particle Swarm Optimization to directly determine the structural parameters of the metasurfaces across the full design space. To validate this approach, we designed and fabricated a line-focusing metalens (LFM) operating at 0.1 THz for linear array detection. Experimental results show a focal spot with a full width at half maximum (FWHM) of 4.1 mm, along with a sidelobe energy of 42%, an energy efficiency of 49%, and a consistent 2 mm resolution maintained over a 20 mm depth of field. When integrated into a line-scanning system, the LFM enables single-pass acquisition of complete images, supporting imaging at 66.7 mm/s while preserving sub-diffraction-limit resolution. This design strategy is expected to provide a reference solution for inverse design in terahertz imaging.</div></div>","PeriodicalId":49719,"journal":{"name":"Optics and Lasers in Engineering","volume":"200 ","pages":"Article 109586"},"PeriodicalIF":3.7,"publicationDate":"2026-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979640","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Hi-ILT: A hybrid End-to-End framework of lightweight hierarchical VAE and physics-Guided ILT for inverse lithography technology Hi-ILT：用于反光刻技术的轻量级分层VAE和物理引导ILT的混合端到端框架

IF 3.7 2区工程技术 Q2 OPTICS

Optics and Lasers in Engineering

Pub Date : 2026-01-13 DOI: 10.1016/j.optlaseng.2026.109621

Hui Xu , Tao Wang , Ziqiang Zhao , Ruijun Ma , Xiaoqing Wen , Huaguo Liang

Optical Proximity Correction (OPC) is a core technology for compensating lithographic diffraction effects in advanced manufacturing processes. Although Inverse Lithography Technology (ILT) enables high-precision mask optimization, it faces challenges in balancing mask printability and optimization efficiency due to three key problems: (1) low-quality initial solutions, (2) high computational overhead, and (3) the inability of pure data-driven methods to accurately model Sub-Resolution Assist Features (SRAF). To address these problems, this paper proposes Hi-ILT, an end-to-end OPC framework that integrates lightweight deep learning and physical ILT correction. A lightweight Convolutional Neural Network (CNN) first generates a high-precision initial mask at low resolution, with a Binary-Straight-Through Estimator (BSTE) resolving binarization gradient vanishing to stabilize training and achieve fast convergence. Subsequently, a gradient descent based ILT correction layer performs fine-grained optimization of mask details (especially SRAF) at high resolution and models process variations. Experiments on 32 nm node M1 layouts (ICCAD 2013 benchmark) and 10 large-scale datasets demonstrate that Hi-ILT outperforms state-of-the-art methods. Compared to deep learning-based ILT methods, it reduces L₂ error by up to 30.1%, Process Variation Band (PVB) by up to 19.8%, and Edge Placement Error (EPE) violations by up to 73.4%; compared to gradient descent-based ILT methods, it shortens end-to-end Turn Around Time (TAT) by up to 69.4% while maintaining higher precision. Hi-ILT effectively achieves a balance between high printability and efficient optimization, making it suitable for advanced lithography requirements.

光学接近校正（OPC）是先进制造工艺中补偿光刻衍射效应的核心技术。虽然逆光刻技术（ILT）可以实现高精度掩模优化，但由于三个关键问题，它在平衡掩模可打印性和优化效率方面面临挑战：(1)低质量的初始解，(2)高计算开销，(3)纯数据驱动方法无法准确建模亚分辨率辅助特征（SRAF）。为了解决这些问题，本文提出了Hi-ILT，这是一个集成了轻量级深度学习和物理ILT校正的端到端OPC框架。轻量级卷积神经网络（CNN）首先在低分辨率下生成高精度初始掩码，利用二值化梯度估计器（BSTE）解决二值化梯度消失，稳定训练，实现快速收敛。随后，基于梯度下降的ILT校正层在高分辨率下对掩模细节（特别是SRAF）进行细粒度优化，并模拟过程变化。在32个 nm节点M1布局（ICCAD 2013基准）和10个大规模数据集上的实验表明，Hi-ILT优于最先进的方法。与基于深度学习的ILT方法相比，它将L2误差降低了30.1%，将过程变化带（PVB）降低了19.8%，将边缘放置误差（EPE）违规降低了73.4%；与基于梯度下降的ILT方法相比，它将端到端周转时间（TAT）缩短了69.4%，同时保持了更高的精度。Hi-ILT有效地实现了高印刷性和高效优化之间的平衡，使其适合高级光刻要求。

{"title":"Hi-ILT: A hybrid End-to-End framework of lightweight hierarchical VAE and physics-Guided ILT for inverse lithography technology","authors":"Hui Xu , Tao Wang , Ziqiang Zhao , Ruijun Ma , Xiaoqing Wen , Huaguo Liang","doi":"10.1016/j.optlaseng.2026.109621","DOIUrl":"10.1016/j.optlaseng.2026.109621","url":null,"abstract":"<div><div>Optical Proximity Correction (OPC) is a core technology for compensating lithographic diffraction effects in advanced manufacturing processes. Although Inverse Lithography Technology (ILT) enables high-precision mask optimization, it faces challenges in balancing mask printability and optimization efficiency due to three key problems: (1) low-quality initial solutions, (2) high computational overhead, and (3) the inability of pure data-driven methods to accurately model Sub-Resolution Assist Features (SRAF). To address these problems, this paper proposes Hi-ILT, an end-to-end OPC framework that integrates lightweight deep learning and physical ILT correction. A lightweight Convolutional Neural Network (CNN) first generates a high-precision initial mask at low resolution, with a Binary-Straight-Through Estimator (BSTE) resolving binarization gradient vanishing to stabilize training and achieve fast convergence. Subsequently, a gradient descent based ILT correction layer performs fine-grained optimization of mask details (especially SRAF) at high resolution and models process variations. Experiments on 32 nm node M1 layouts (ICCAD 2013 benchmark) and 10 large-scale datasets demonstrate that Hi-ILT outperforms state-of-the-art methods. Compared to deep learning-based ILT methods, it reduces <em>L</em><sub>2</sub> error by up to 30.1%, Process Variation Band (PVB) by up to 19.8%, and Edge Placement Error (EPE) violations by up to 73.4%; compared to gradient descent-based ILT methods, it shortens end-to-end Turn Around Time (TAT) by up to 69.4% while maintaining higher precision. Hi-ILT effectively achieves a balance between high printability and efficient optimization, making it suitable for advanced lithography requirements.</div></div>","PeriodicalId":49719,"journal":{"name":"Optics and Lasers in Engineering","volume":"200 ","pages":"Article 109621"},"PeriodicalIF":3.7,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979643","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Space debris detection method for sequential optical astronomical images 序列光学天文图像空间碎片检测方法

IF 3.7 2区工程技术 Q2 OPTICS

Optics and Lasers in Engineering

Pub Date : 2026-01-13 DOI: 10.1016/j.optlaseng.2025.109598

Lanbo Zhao, Ning Yang, Jun Fang, Jie Han, Wei Wei, Lin Hu

The detection of space debris plays a crucial role in ensuring the safety of satellites and spacecraft in Earth orbit. However, due to significant background noise and a low signal-to-noise ratio (SNR) in optical astronomical images, accurately detecting dim and small debris targets remains a major challenge. This paper proposes a method for detecting space debris in optical image sequences. The method first applies an improved adaptive filtering algorithm to precisely suppress background noise while preserving dim targets. Next, we integrate feature detection with optical-flow analysis to capture inter-frame motion information, and extract candidate targets using the relative inter-frame motion distance (RIMD). Finally, targets are validated via a spatiotemporal association model with trajectory compensation to reduce missed detections and improve accuracy. On the SpotGEO benchmark, our method achieves a detection rate of 95.09% with a false alarm rate of 6.02% and an average runtime of 1.06 s per sequence; on real astronomical observations, it attains a mean sub-pixel centroid-localization accuracy of 0.4608 px. Experimental results show that the proposed method has broad applicability and excellent real-time performance, providing a reliable solution for space debris detection.

空间碎片探测对确保地球轨道卫星和航天器的安全具有至关重要的作用。然而，由于光学天文图像中存在较大的背景噪声和较低的信噪比，因此精确检测弱小碎片目标仍然是一个重大挑战。提出了一种基于光学图像序列的空间碎片检测方法。该方法首先采用一种改进的自适应滤波算法，在保持弱小目标的同时精确抑制背景噪声。接下来，我们将特征检测与光流分析相结合，捕捉帧间运动信息，并利用相对帧间运动距离（RIMD）提取候选目标。最后，通过具有轨迹补偿的时空关联模型对目标进行验证，以减少漏检，提高检测精度。在SpotGEO基准测试中，我们的方法检测率为95.09%，虚警率为6.02%，每个序列的平均运行时间为1.06 s；在实际天文观测中，平均亚像素质心定位精度为0.4608 px。实验结果表明，该方法具有广泛的适用性和优异的实时性，为空间碎片检测提供了可靠的解决方案。

{"title":"Space debris detection method for sequential optical astronomical images","authors":"Lanbo Zhao, Ning Yang, Jun Fang, Jie Han, Wei Wei, Lin Hu","doi":"10.1016/j.optlaseng.2025.109598","DOIUrl":"10.1016/j.optlaseng.2025.109598","url":null,"abstract":"<div><div>The detection of space debris plays a crucial role in ensuring the safety of satellites and spacecraft in Earth orbit. However, due to significant background noise and a low signal-to-noise ratio (SNR) in optical astronomical images, accurately detecting dim and small debris targets remains a major challenge. This paper proposes a method for detecting space debris in optical image sequences. The method first applies an improved adaptive filtering algorithm to precisely suppress background noise while preserving dim targets. Next, we integrate feature detection with optical-flow analysis to capture inter-frame motion information, and extract candidate targets using the relative inter-frame motion distance (RIMD). Finally, targets are validated via a spatiotemporal association model with trajectory compensation to reduce missed detections and improve accuracy. On the SpotGEO benchmark, our method achieves a detection rate of 95.09% with a false alarm rate of 6.02% and an average runtime of 1.06 s per sequence; on real astronomical observations, it attains a mean sub-pixel centroid-localization accuracy of 0.4608 px. Experimental results show that the proposed method has broad applicability and excellent real-time performance, providing a reliable solution for space debris detection.</div></div>","PeriodicalId":49719,"journal":{"name":"Optics and Lasers in Engineering","volume":"200 ","pages":"Article 109598"},"PeriodicalIF":3.7,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979645","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An improved PatchMatch-based stereo matching method for occlusion and boundary handling 一种改进的基于patchmatch的遮挡和边界立体匹配方法

IF 3.7 2区工程技术 Q2 OPTICS

Optics and Lasers in Engineering

Pub Date : 2026-01-13 DOI: 10.1016/j.optlaseng.2026.109622

Xiang Xiong , Yibo Li , Liying Sun , Qian Liu

Stereo matching is a promising technology for three-dimensional measurement, yet it remains challenged by occluded regions and disparity boundaries. Researchers have explored various strategies, among which PatchMatch-based methods have become prominent due to their efficiency and non-local processing capabilities. To tackle these issues, we propose an innovative approach that employs a novel disparity continuity definition to standardize the smoothness loss at disparity boundaries. Furthermore, by integrating graph cut and local expansion move methods, we achieve reliable disparity estimation for occluded regions. Additionally, to enhance computational efficiency, we implement superpixel-guided filtering technology to constrain the propagation of erroneous disparities. The proposed algorithm is evaluated on two authoritative benchmarks: Middlebury V2 and V3. It achieves an average bad1.0 metric of 6.84 % on V2 and 12.8 % on V3, outperforming existing PatchMatch-based methods in both accuracy and efficiency. Furthermore, we validate the generalization capability of our approach through comparisons with deep learning-based methods. We hope our method will inspire future research aimed at addressing the bottleneck problems in stereo matching.

立体匹配是一种很有前途的三维测量技术，但它仍然受到遮挡区域和视差边界的挑战。研究人员探索了各种策略，其中基于patchmatch的方法因其高效和非局部处理能力而备受关注。为了解决这些问题，我们提出了一种创新的方法，采用新的视差连续性定义来标准化视差边界的平滑损失。结合图切和局部展开移动方法，实现了对遮挡区域的可靠视差估计。此外，为了提高计算效率，我们实现了超像素引导滤波技术来限制错误差的传播。该算法在两个权威基准上进行了评估：Middlebury V2和V3。它在V2和V3上的平均bad1.0指标分别达到6.84%和12.8%，在准确性和效率方面都优于现有的基于patchmatch的方法。此外，我们通过与基于深度学习的方法的比较来验证我们的方法的泛化能力。我们希望我们的方法能够启发未来的研究，旨在解决立体匹配中的瓶颈问题。

{"title":"An improved PatchMatch-based stereo matching method for occlusion and boundary handling","authors":"Xiang Xiong , Yibo Li , Liying Sun , Qian Liu","doi":"10.1016/j.optlaseng.2026.109622","DOIUrl":"10.1016/j.optlaseng.2026.109622","url":null,"abstract":"<div><div>Stereo matching is a promising technology for three-dimensional measurement, yet it remains challenged by occluded regions and disparity boundaries. Researchers have explored various strategies, among which PatchMatch-based methods have become prominent due to their efficiency and non-local processing capabilities. To tackle these issues, we propose an innovative approach that employs a novel disparity continuity definition to standardize the smoothness loss at disparity boundaries. Furthermore, by integrating graph cut and local expansion move methods, we achieve reliable disparity estimation for occluded regions. Additionally, to enhance computational efficiency, we implement superpixel-guided filtering technology to constrain the propagation of erroneous disparities. The proposed algorithm is evaluated on two authoritative benchmarks: Middlebury V2 and V3. It achieves an average bad1.0 metric of 6.84 % on V2 and 12.8 % on V3, outperforming existing PatchMatch-based methods in both accuracy and efficiency. Furthermore, we validate the generalization capability of our approach through comparisons with deep learning-based methods. We hope our method will inspire future research aimed at addressing the bottleneck problems in stereo matching.</div></div>","PeriodicalId":49719,"journal":{"name":"Optics and Lasers in Engineering","volume":"200 ","pages":"Article 109622"},"PeriodicalIF":3.7,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979642","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A cross-cavity dual-crystal four-frequency tunable laser with orthogonal polarization and wide frequency-difference tuning 一种正交偏振宽频差调谐的跨腔双晶四频可调谐激光器

IF 3.7 2区工程技术 Q2 OPTICS

Optics and Lasers in Engineering

Pub Date : 2026-01-12 DOI: 10.1016/j.optlaseng.2026.109619

Junhong Xing, Zhou Zheng, Jinrong Xu, Wenyuan Liu, Lei Xi, Mingxing Jiao, Yun Liu

We demonstrate a tunable four-frequency laser system based on a cross-cavity design with dual Yb:YAG crystals. Each crystal generates two orthogonally polarized beams at 1030 nm from two independent cavities formed by a polarized beam splitter (PBS). By employing birefringent half-wave plates (HWPs) as tuning elements, single-longitudinal-mode oscillation is selectively achieved in each cavity. The system emits four output frequencies with orthogonal polarizations. Frequency difference tuning is accomplished via angular adjustment of the HWPs, achieving a maximum frequency difference of 1.02 THz. The output power of each frequency can be balanced through independent control of cavity losses. This work provides a compact and stable platform for multi-frequency laser generation, with promising applications in spectral detection, metrology, and terahertz wave generation.

我们展示了一种基于双Yb:YAG晶体交叉腔设计的可调谐四频激光系统。每个晶体产生两个正交极化光束在1030nm从两个独立的腔由极化分束器（PBS）形成。采用双折射半波片（HWPs）作为调谐元件，可选择性地在每个腔内实现单纵模振荡。该系统发射四个正交极化的输出频率。频率差调谐是通过HWPs的角度调整来完成的，最大频率差为1.02太赫兹。每个频率的输出功率可以通过独立控制腔损耗来平衡。这项工作为多频激光的产生提供了一个紧凑而稳定的平台，在光谱检测、计量和太赫兹波的产生方面具有广阔的应用前景。

引用次数: 0

Single -shot imaging through scattering media and around the corner beyond the OME range via polarization-encoded spatial multiplexing 通过偏振编码的空间多路复用，通过散射介质和在OME范围以外的角落进行单次成像

IF 3.7 2区工程技术 Q2 OPTICS

Optics and Lasers in Engineering

Pub Date : 2026-01-12 DOI: 10.1016/j.optlaseng.2026.109602

Yi Wei , Yan Zhao , Taotao Qin , Lingfeng Liu , Lianfa Bai , Jing Han , Yingjie Shi , Enlai Guo

Imaging through scattering media and around corners faces fundamental field-of-view limitations imposed by the optical memory effect (OME), where single-shot solutions remain scarce. To overcome these barriers, this work introduces a polarization-encoded spatial multiplexing method that achieves beyond-OME reconstructions within single shots. By establishing a linear polarization mapping between detected speckles and polarization-specific speckles across distinct OME regions, we develop a speckle demultiplexing framework to extract speckles across distinct OME ranges for respective reconstructions. They can be precisely achieved by the designed two-step algorithm, incorporating the N-FINDR estimation and the truncated Cauchy non-negative matrix factorization with the local neighborhood weights and the graph Laplacian constraints. Validated in transmissive/reflective systems, the method achieves 4.5× FOV expansion with 32 dB speckle demultiplexing fidelity, enabling satisfying hidden scene recovery. Its single-shot beyond-OME imaging capability shows promising application potential in fast imaging.

通过散射介质和拐角进行成像面临着由光记忆效应（OME）施加的基本视场限制，其中单镜头解决方案仍然很少。为了克服这些障碍，本研究引入了一种偏振编码的空间复用方法，该方法可以在单次拍摄中实现超ome重建。通过在检测到的散斑和不同OME区域的偏振特异性散斑之间建立线性偏振映射，我们开发了一个散斑解复用框架，以提取不同OME范围的散斑进行各自的重建。通过结合N-FINDR估计和截断柯西非负矩阵分解与局部邻域权值和图拉普拉斯约束的两步算法，可以精确地实现这些目标。在透射/反射系统中验证，该方法实现了4.5× FOV扩展，32 dB散斑解复用保真度，实现了令人满意的隐藏场景恢复。它的单次超ome成像能力在快速成像中显示出很大的应用潜力。

{"title":"Single -shot imaging through scattering media and around the corner beyond the OME range via polarization-encoded spatial multiplexing","authors":"Yi Wei , Yan Zhao , Taotao Qin , Lingfeng Liu , Lianfa Bai , Jing Han , Yingjie Shi , Enlai Guo","doi":"10.1016/j.optlaseng.2026.109602","DOIUrl":"10.1016/j.optlaseng.2026.109602","url":null,"abstract":"<div><div>Imaging through scattering media and around corners faces fundamental field-of-view limitations imposed by the optical memory effect (OME), where single-shot solutions remain scarce. To overcome these barriers, this work introduces a polarization-encoded spatial multiplexing method that achieves beyond-OME reconstructions within single shots. By establishing a linear polarization mapping between detected speckles and polarization-specific speckles across distinct OME regions, we develop a speckle demultiplexing framework to extract speckles across distinct OME ranges for respective reconstructions. They can be precisely achieved by the designed two-step algorithm, incorporating the N-FINDR estimation and the truncated Cauchy non-negative matrix factorization with the local neighborhood weights and the graph Laplacian constraints. Validated in transmissive/reflective systems, the method achieves 4.5× FOV expansion with 32 dB speckle demultiplexing fidelity, enabling satisfying hidden scene recovery. Its single-shot beyond-OME imaging capability shows promising application potential in fast imaging.</div></div>","PeriodicalId":49719,"journal":{"name":"Optics and Lasers in Engineering","volume":"200 ","pages":"Article 109602"},"PeriodicalIF":3.7,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979638","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Efficient dual-domain iterative algorithm for computed laminography 计算机层析成像的高效双域迭代算法

IF 3.7 2区工程技术 Q2 OPTICS

Optics and Lasers in Engineering

Pub Date : 2026-01-10 DOI: 10.1016/j.optlaseng.2025.109590

Pengxiang Ji, Tian He, Haihang Zhao, Jing Zou

X-ray computed laminography (CL) has emerged as a highly effective technique for non-destructive testing of plate-like samples. However, it is challenging to strike an optimal balance between image quality and computation speed due to the intractable superimposed artifacts. In this paper, a novel strategy based on Fourier information completeness is developed for high-speed de-artifacting reconstruction of CL. We first demonstrate that the Fourier defective volume of CL is delineated by multiple ellipses that share a common focus. Guided by this geometric insight, a resampling algorithm is designed to maximally restore the Fourier information of cone-beam CL projections. The restored data is then incorporated into an efficient dual-domain framework, and complemented by L₀ norm regularization in image domain. Numerical experiments demonstrate that the proposed method has achieved not only comparable image quality to that of conventional SART-based algorithm, but also an efficiency level approximately 32 times higher. The proposed elliptical model and resampling algorithm provide a new interpretation of cone-beam CL data, and hold great promise for real-time reconstruction in industrial applications.

x射线计算机层析成像（CL）已成为板状样品无损检测的一种高效技术。然而，由于难以处理的叠加伪影，在图像质量和计算速度之间取得最佳平衡是一项挑战。本文提出了一种基于傅里叶信息完备性的高速去伪重建策略。我们首先证明了CL的傅立叶缺陷体积是由多个具有共同焦点的椭圆所描绘的。在这种几何洞察力的指导下，设计了一种重采样算法，以最大限度地恢复锥束CL投影的傅里叶信息。然后将恢复后的数据整合到有效的双域框架中，并在图像域进行L0范数正则化。数值实验表明，该方法不仅获得了与传统基于sart算法相当的图像质量，而且效率提高了约32倍。所提出的椭圆模型和重采样算法为锥束CL数据提供了新的解释，在工业应用中的实时重建具有很大的前景。

{"title":"Efficient dual-domain iterative algorithm for computed laminography","authors":"Pengxiang Ji, Tian He, Haihang Zhao, Jing Zou","doi":"10.1016/j.optlaseng.2025.109590","DOIUrl":"10.1016/j.optlaseng.2025.109590","url":null,"abstract":"<div><div>X-ray computed laminography (CL) has emerged as a highly effective technique for non-destructive testing of plate-like samples. However, it is challenging to strike an optimal balance between image quality and computation speed due to the intractable superimposed artifacts. In this paper, a novel strategy based on Fourier information completeness is developed for high-speed de-artifacting reconstruction of CL. We first demonstrate that the Fourier defective volume of CL is delineated by multiple ellipses that share a common focus. Guided by this geometric insight, a resampling algorithm is designed to maximally restore the Fourier information of cone-beam CL projections. The restored data is then incorporated into an efficient dual-domain framework, and complemented by L<sub>0</sub> norm regularization in image domain. Numerical experiments demonstrate that the proposed method has achieved not only comparable image quality to that of conventional SART-based algorithm, but also an efficiency level approximately 32 times higher. The proposed elliptical model and resampling algorithm provide a new interpretation of cone-beam CL data, and hold great promise for real-time reconstruction in industrial applications.</div></div>","PeriodicalId":49719,"journal":{"name":"Optics and Lasers in Engineering","volume":"200 ","pages":"Article 109590"},"PeriodicalIF":3.7,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145928042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0