Infrared Physics & Technology最新文献_第7页

Multivariate calibration models for ATR-FTIR spectroscopy analysis of olive oil adulterated with corn, sunflower, and palm oils 掺入玉米、向日葵和棕榈油的橄榄油的ATR-FTIR光谱分析的多变量校准模型

IF 3.4 3区物理与天体物理

Infrared Physics & Technology Pub Date : 2026-03-01 Epub Date: 2026-01-21 DOI: 10.1016/j.infrared.2026.106420

Esra modares Askari , Mohammadreza Khanmohammadi Khorrami , Mahsa Mohammadi , Miguel de la Guardia

{"title":"Multivariate calibration models for ATR-FTIR spectroscopy analysis of olive oil adulterated with corn, sunflower, and palm oils","authors":"Esra modares Askari , Mohammadreza Khanmohammadi Khorrami , Mahsa Mohammadi , Miguel de la Guardia","doi":"10.1016/j.infrared.2026.106420","DOIUrl":"10.1016/j.infrared.2026.106420","url":null,"abstract":"<div><div>Rapid analysis using attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy has been developed for the detection of extra virgin olive oil (EVOO) adulterated with corn oil, sunflower oil (SFO), and palm oil. Measurements were conducted on EVOO and samples adulterated with varying contents of oils, ranging from 1.0 to 35% by weight. ATR-FTIR spectra were analyzed using a robust principal component analysis–multivariate adaptive regression splines (rPCA-MARS) model to assess the adulterated samples. Artificial neural networks (ANN) and support vector machines regression (SVM-R) were used to predict olive oil adulteration. The performance of these models, along with linear and cubic rPCA-MARS models, was evaluated using R<sup>2</sup> and RMSE. The linear rPCA-MARS model demonstrated the highest accuracy in predicting adulteration, with an R<sup>2</sup>pred value of 0.970 and MSE (pred<strong>)</strong> of 4.950. This model outperformed ANN, SVM-R, and piecewise cubic rPCA-MARS models based on key statistical metrics.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"154 ","pages":"Article 106420"},"PeriodicalIF":3.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146074074","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Intelligent origin discrimination of Wuyi Rock Tea during storage using a time-spectral dual-dimensional transformer model 基于时间谱二维变压器模型的武夷岩茶储存期产地智能判别

IF 3.4 3区物理与天体物理

Infrared Physics & Technology Pub Date : 2026-03-01 Epub Date: 2026-01-27 DOI: 10.1016/j.infrared.2026.106411

Xue Bai , Min Sun , Xianshu Fu , Jinyan Liao , Mingzhou Zhang , Xiaoping Yu , Zihong Ye

{"title":"Intelligent origin discrimination of Wuyi Rock Tea during storage using a time-spectral dual-dimensional transformer model","authors":"Xue Bai , Min Sun , Xianshu Fu , Jinyan Liao , Mingzhou Zhang , Xiaoping Yu , Zihong Ye","doi":"10.1016/j.infrared.2026.106411","DOIUrl":"10.1016/j.infrared.2026.106411","url":null,"abstract":"<div><div>This study addresses the challenge of accurately discriminating the origin of Wuyi Rock Tea due to dynamic changes in its compound composition during storage. A novel intelligent discrimination method based on a time-spectral dual-dimensional dynamic model is proposed. By integrating Near-Infrared (NIR) spectroscopy with machine learning techniques, Principal Component Analysis (PCA) combined with Isolation Forest (iForest) was employed to eliminate anomalous samples, while 12 single preprocessing methods and 6 hybrid combinations were introduced to enhance data quality. The performance of four traceability models—K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Random Forest (RF), and Transformer—was evaluated, with a focus on optimizing the Transformer model through input dimension screening. The term ’time-spectral dual-dimensional’ denotes the fusion of storage time and spectral (NIR) data via the Transformer’s self-attention mechanism, capturing spectral evolution over time without employing time-series architectures. Results demonstrated that the optimized Transformer model combined with first-order derivative (1D) preprocessing (O-1D-Transformer) achieved the highest performance, with test set accuracy, Macro-F1, and G-mean values of 98.05%, 97.11%, and 98.41%, respectively. By dynamically modulating inter-band correlations via self-attention, this model captures storage-induced chemical patterns, distinguishing origin signals from chronological noise and thereby overcoming the time-blindness of static models. This research provides a new paradigm for intelligent traceability of time-sensitive food products, validating the synergistic advantages of NIR spectroscopy and deep learning, with significant implications for tea quality regulation and brand protection.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"154 ","pages":"Article 106411"},"PeriodicalIF":3.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147396831","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Quantile interval prediction strategy enhances quantitative characterization of mycotoxin content in wheat using Near-Infrared spectroscopy 分位区间预测策略增强了小麦真菌毒素含量的近红外光谱定量表征

IF 3.4 3区物理与天体物理

Infrared Physics & Technology Pub Date : 2026-03-01 Epub Date: 2026-01-31 DOI: 10.1016/j.infrared.2026.106441

Jingwen Zhu , Xuehao Shen , Xianjun Sun , Huazhi Wang , Hui Jiang

{"title":"Quantile interval prediction strategy enhances quantitative characterization of mycotoxin content in wheat using Near-Infrared spectroscopy","authors":"Jingwen Zhu , Xuehao Shen , Xianjun Sun , Huazhi Wang , Hui Jiang","doi":"10.1016/j.infrared.2026.106441","DOIUrl":"10.1016/j.infrared.2026.106441","url":null,"abstract":"<div><div>Rapid and reliable detection of mycotoxins in wheat is of great significance for ensuring food safety and risk warning. This study combines portable near-infrared spectroscopy technology with quantile regression deep learning (QRDL) to construct an efficient and information-rich method for predicting the range of mycotoxin content. This study compared classification models and point prediction regression models using convolutional neural networks (CNN) and long short-term memory (LSTM), as well as interval prediction models based on quantile regression convolutional neural networks (QR-CNN) and quantile regression long short-term memory (QR-LSTM). The results show that the interval prediction model demonstrates excellent performance in terms of interval width control, median fitting accuracy, and control of the number of samples with inter-class bias. It effectively enhances the ability to express uncertainties, and the interpretability of predictions. This study provides more robust information support for the identification of mycotoxin risk levels and on-site rapid detection in wheat.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"154 ","pages":"Article 106441"},"PeriodicalIF":3.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147397540","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Intracavity switchable Laguerre–Gaussian mode generation in a side-pumped Nd:YAG laser via spot defect and off-axis pumping 利用光斑缺陷和离轴泵浦产生侧泵浦Nd:YAG激光器腔内可切换拉盖尔-高斯模式

IF 3.4 3区物理与天体物理

Infrared Physics & Technology Pub Date : 2026-03-01 Epub Date: 2026-01-01 DOI: 10.1016/j.infrared.2026.106366

Zeqian Liu , Hui Chen , Jiashuo An , Junhong Chen , Bingzheng Yan , Yulei Wang , Zhiwei Lu , Zhenxu Bai

{"title":"Intracavity switchable Laguerre–Gaussian mode generation in a side-pumped Nd:YAG laser via spot defect and off-axis pumping","authors":"Zeqian Liu , Hui Chen , Jiashuo An , Junhong Chen , Bingzheng Yan , Yulei Wang , Zhiwei Lu , Zhenxu Bai","doi":"10.1016/j.infrared.2026.106366","DOIUrl":"10.1016/j.infrared.2026.106366","url":null,"abstract":"<div><div>Laguerre-Gaussian (LG) beams carrying orbital angular momentum (OAM) demonstrate enhanced information-carrying capacity, making them particularly valuable for high-capacity optical communications. Additionally, petal-like LG beams exhibit unique advantages for spin angular velocity measurements due to their distinctive intensity profiles and phase singularity characteristics. However, current laser systems face significant challenges in achieving direct switching between high-order vortex beams and petal-like beams outputs. This study presents a side-pumped Nd:YAGlaser system capable of direct switching between high-order vortex beams and petal-like beams. By introducing specifically engineered spot defects of varying dimensions in the resonator mirrors of a side-pumped Nd:YAGlaser system, we demonstrate tunable vortex beam generation from 1st to 6th order through cavity length adjustments. Furthermore, mode conversion between vortex and petal-like beams is achieved using an off-axis pumping scheme, in which the pump beam is intentionally displaced from the center of the defect. Using this approach, stable petal-like LG modes ranging from LG<sub>0,1</sub> to LG<sub>0,12</sub>. are successfully generated. To the best of our knowledge, this work demonstrates for the first time an intracavity LG mode conversion strategy that combines engineered spot defects with off-axis pumping, providing a compact and versatile platform for flexible structured-light generation.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"154 ","pages":"Article 106366"},"PeriodicalIF":3.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145923919","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A novel diffusion-based background estimation for infrared dim small target detection 一种新的基于扩散的红外弱小目标检测背景估计

IF 3.4 3区物理与天体物理

Infrared Physics & Technology Pub Date : 2026-03-01 Epub Date: 2026-01-10 DOI: 10.1016/j.infrared.2026.106384

Sheng-hui Rong , Wang Zi-ming , Gao Xue-zhen , Zhao Wen-feng , Wu Xu-peng , Zhang Tao

{"title":"A novel diffusion-based background estimation for infrared dim small target detection","authors":"Sheng-hui Rong , Wang Zi-ming , Gao Xue-zhen , Zhao Wen-feng , Wu Xu-peng , Zhang Tao","doi":"10.1016/j.infrared.2026.106384","DOIUrl":"10.1016/j.infrared.2026.106384","url":null,"abstract":"<div><div>Infrared small target detection is a crucial technique in the field of computer vision. With the advancement of deep learning, Convolutional Neural Network (CNN)-based methods have achieved promising results in target detection. However, due to the small size of the targets, relying solely on dense convolutional layers may lead to information loss. To address the issue of inaccurate background prediction in complex background, we propose an end-to-end infrared background prediction method based on a conditional diffusion model with an adaptive blocking strategy. On one hand, the adaptive blocking strategy effectively integrates both local and global information from the infrared image while significantly accelerating the inference speed of the diffusion model. On the other hand, the multi-scale attention segmentation module can effectively suppress background clutter and enhance the target. We also created an IRDF (infrared for diffusion) dataset, comprising of 23,378 images to evaluate the detection performance of the proposed method and the comparison methods. Extensive experiments demonstrate that our approach is capable of detecting targets precisely and performs effectively in various complex backgrounds.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"154 ","pages":"Article 106384"},"PeriodicalIF":3.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023730","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

HR-YOLO: Enhanced YOLO framework for infrared-based timber component damage detection in heritage arch lounge bridges HR-YOLO：基于红外的传统拱廊桥木材构件损伤检测的增强YOLO框架

IF 3.4 3区物理与天体物理

Infrared Physics & Technology Pub Date : 2026-03-01 Epub Date: 2026-01-14 DOI: 10.1016/j.infrared.2026.106401

Zhiqiang Qin , Qifang Xie , Zheng Li

{"title":"HR-YOLO: Enhanced YOLO framework for infrared-based timber component damage detection in heritage arch lounge bridges","authors":"Zhiqiang Qin , Qifang Xie , Zheng Li","doi":"10.1016/j.infrared.2026.106401","DOIUrl":"10.1016/j.infrared.2026.106401","url":null,"abstract":"<div><div>This study proposes an infrared imaging-based damage detection method for ancient Chinese timber arch lounge bridges using an improved YOLO11 model. Three datasets were established for damage detection scenarios involving general timber components and heritage timber arch lounge bridge components. Although the YOLO11 model achieved good performance after hyperparameter optimization in general timber components detection task (69.3% on cracks; 63.2% on multi-damage), its precision dropped significantly in heritage timber arch lounge bridge scenarios due to small targets and complex backgrounds. A transfer learning strategy was adopted to overcome this limitation, which enhanced training stability and detection precision for heritage bridge damage detection. Moreover, Non-local Attention Mechanism and SimAM were integrated into the YOLO11 architecture to improve the model performance over three datasets. Detection results demonstrated that the Non-local Attention Mechanism achieves higher precision and is more suitable for infrared video detection tasks compared to the baseline model. SimAM exhibits significant lightweight characteristics while maintaining great precision. Finally, we outline a practical UAV-based inspection workflow that couples infrared data acquisition with real-time HR-YOLO inference to support repeatable, field-ready monitoring of heritage timber bridges. This research offers an automated solution for real-time structural health monitoring of cultural heritage, promoting the application of intelligent infrared detection techniques in heritage preservation and civil engineering.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"154 ","pages":"Article 106401"},"PeriodicalIF":3.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023731","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multimodal online monitoring of laser melt pool based on polarized infrared spectral imaging 基于偏振红外光谱成像的激光熔池多模态在线监测

IF 3.4 3区物理与天体物理

Infrared Physics & Technology Pub Date : 2026-03-01 Epub Date: 2026-01-14 DOI: 10.1016/j.infrared.2026.106404

Xianjin Kong, Guibo Chen, Ye Zhang

{"title":"Multimodal online monitoring of laser melt pool based on polarized infrared spectral imaging","authors":"Xianjin Kong, Guibo Chen, Ye Zhang","doi":"10.1016/j.infrared.2026.106404","DOIUrl":"10.1016/j.infrared.2026.106404","url":null,"abstract":"<div><div>This study proposes a multimodal diagnostic approach that integrates infrared polarization spectroscopy imaging, spectral temperature inversion, and numerical simulation to investigate the dynamic evolution of melt pool behavior during laser processing. A near-infrared polarization imaging system was employed to achieve real-time observation of the melt pool interface and morphological evolution. By incorporating the multispectral thermometry (MST) method, temperature reconstruction was conducted, revealing a strong spatial correlation between the saturated regions in the degree of linear polarization (DOLP) images and localized high-temperature zones. To address the limitation of polarization imaging in capturing melt pool depth, a multiphysics model incorporating key mechanical mechanisms was established to analyze the coupled evolution of temperature and flow fields and to support interface morphology interpretation. Experimental results demonstrate that the interface contours extracted from DOLP images closely match the boundaries observed in thermal imaging, with deviations between experimental observations and numerical predictions remaining below 10%. In the quasi-steady-state regime, interfacial driving force analysis indicates that recoil pressure predominantly governs the formation of interfacial depressions, while Marangoni forces control the recirculation of molten metal and the flattening of edge morphology. Furthermore, the study reveals nonlinear thermal and morphological responses of the melt pool under varying combinations of pulse width and repetition rate. The integrated experimental–numerical framework developed in this work enables melt-pool behavior identification and mechanism interpretation, and provides a foundation for predictive analysis of interface evolution, offering a systematic basis for understanding interfacial dynamics in laser processing.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"154 ","pages":"Article 106404"},"PeriodicalIF":3.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023739","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

DTRDM: Dual-task directional residual denoising diffusion model for multimodal image fusion 多模态图像融合的双任务定向残差去噪扩散模型

IF 3.4 3区物理与天体物理

Infrared Physics & Technology Pub Date : 2026-03-01 Epub Date: 2026-01-07 DOI: 10.1016/j.infrared.2026.106389

Jinlin Jiang , Gang Hu , Guanglei Sheng , Guo Wei

{"title":"DTRDM: Dual-task directional residual denoising diffusion model for multimodal image fusion","authors":"Jinlin Jiang , Gang Hu , Guanglei Sheng , Guo Wei","doi":"10.1016/j.infrared.2026.106389","DOIUrl":"10.1016/j.infrared.2026.106389","url":null,"abstract":"<div><div>Image fusion enhances complementary details and visual quality by integrating information from multiple modalities, thereby supporting more accurate decision-making in downstream tasks. While diffusion models show strong generative ability in fusion tasks, the absence of real-image supervision restricts their ability to capture local features. To address this, we propose a Dual-Task Directed Residual Denoising Diffusion Model (DTRDM) to better capture multi-scale diffusion features and enrich fused image content. First, we introduce two diffusion biases: “image residuals and pure noise” to guide forward diffusion in a goal-oriented manner. This strategy explicitly guides the inverse fusion process while reducing training complexity. Second, we design a noise prediction module based on a dual U-Net architecture, which generates residual or noise prediction features depending on the training objective. Multi-scale features are refined through cascading and iterative extraction, enabling the model to capture local details across modalities and enhance the fused representation. Finally, we introduce a color–structure-preserving composite loss for denoising, which strengthens feature complementarity across scales. Extensive experiments show that DTRDM achieves state-of-the-art results across key metrics with strong adaptability. It generalizes to diverse fusion tasks without retraining, and its results substantially improve decision-making in applications such as autonomous driving, traffic monitoring, and medical imaging.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"154 ","pages":"Article 106389"},"PeriodicalIF":3.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145974747","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

HFFN: Hierarchical feature fusion network for RGB-infrared object detection HFFN: rgb -红外目标检测的分层特征融合网络

IF 3.4 3区物理与天体物理

Infrared Physics & Technology Pub Date : 2026-03-01 Epub Date: 2026-01-26 DOI: 10.1016/j.infrared.2026.106429

Jinhang Zhang, Min Gao, Yi Wang, Chaowang Li, Dan Fang, Zhuang Wei

{"title":"HFFN: Hierarchical feature fusion network for RGB-infrared object detection","authors":"Jinhang Zhang, Min Gao, Yi Wang, Chaowang Li, Dan Fang, Zhuang Wei","doi":"10.1016/j.infrared.2026.106429","DOIUrl":"10.1016/j.infrared.2026.106429","url":null,"abstract":"<div><div>In recent years, multimodal object detection techniques based on visible(RGB) and infrared(IR) images have gained attention as a research hotspot. These techniques leverage the complementary characteristics of RGB and IR images to overcome the environmental adaptability limitations of single-modality methods. The fusion of RGB and IR images is crucial for object detection under complex lighting conditions. By effectively integrating the complementary information of the two modality images, the robustness and accuracy of detection are enhanced. However, current image fusion methods often encounter issues of inaccurate modality alignment and insufficient representation of fused features during feature extraction and fusion. We propose a Hierarchical Feature Fusion Network (HFFN) that achieves efficient extraction and accurate alignment of multimodal features using an Adaptive Pooling Transformer (APT) and an Invertible Detail Extraction (IDE), significantly enhancing the fusion quality. The APT combines pooling operations and the Transformer structure, enabling it to capture global information and expand the receptive field of features; The IDE focuses on extracting local detail features, including edges and texture information, to enhance the fineness and clarity of the fused image. We constructed object detectors based on the HFFN to validate its effectiveness. Extensive experiments on datasets demonstrate the superior performance of our method.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"154 ","pages":"Article 106429"},"PeriodicalIF":3.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146074069","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

SamFusion:A model for multimodal image fusion guided by SAM’s rich semantics SamFusion：基于SAM丰富语义的多模态图像融合模型

IF 3.4 3区物理与天体物理

Infrared Physics & Technology Pub Date : 2026-03-01 Epub Date: 2026-01-31 DOI: 10.1016/j.infrared.2026.106438

Yucheng Zhang, You Ma, Lin Chai

{"title":"SamFusion:A model for multimodal image fusion guided by SAM’s rich semantics","authors":"Yucheng Zhang, You Ma, Lin Chai","doi":"10.1016/j.infrared.2026.106438","DOIUrl":"10.1016/j.infrared.2026.106438","url":null,"abstract":"<div><div>Infrared and visible image fusion aims to combine the advantages of each modality, such as salient targets in infrared images and detailed textures in visible images, in order to generate images with better visual effects and richer information. Existing fusion methods do not distinguish between semantic and texture information, ignoring their specific roles in the image fusion process, whereas we have demonstrated the critical role of semantic information in the image fusion task. However, the semantic information extraction networks of current methods are restricted to be trained on limited infrared and visible datasets, which results in significantly inadequate semantic information extraction capabilities when facing complex scenes. To address this challenge, we pioneered SamFusion, an image fusion model that leverages the SAM large model, which effectively exploits the rich semantic information in SAM to facilitate image fusion. Specifically, we introduce the Multi Level Semantic Aggregation (MLSA) and Soft Interaction (SI) modules, which aim to mine the semantic information at each level in SAM that is conducive to fusion, while reducing the modality differences in semantic information. In addition, specialized fusion modules are designed for semantic information and texture information respectively. For texture information with significant modality differences, semantic information is used as a guide for selective fusion. Extensive experiments show that our method outperforms the state-of-the-art methods in highlighting salient targets and detailed textures. In addition, the effectiveness of our method is revealed in medical image fusion and downstream tasks of object detection and semantic segmentation.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"154 ","pages":"Article 106438"},"PeriodicalIF":3.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147397541","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0