Infrared Physics & Technology最新文献

Breaking dimensional barriers in hyperspectral target detection: Atrous convolution with Gramian Angular field representations 打破高光谱目标检测中的维度障碍：阿特罗斯卷积与格拉米安角场表示法

IF 3.1 3区物理与天体物理 Q2 INSTRUMENTS & INSTRUMENTATION

Infrared Physics & Technology

Pub Date : 2024-11-13 DOI: 10.1016/j.infrared.2024.105623

Hongzhou Wang , Yulei Wang , Yuchao Yang , Enyu Zhao , Jian Zeng

Hyperspectral images contain extensive spectral bands with rich spectral information that reflects object properties. Leveraging state-of-the-art deep learning techniques has proven to be effective in hyperspectral target detection. However, compared to two-dimensional matrix data, the one-dimensional nature of spectral sequence limits the information that can be extracted, posing a challenge for deep learning-based hyperspectral target detection methodologies. To address this issue, a novel hyperspectral target detection method employing atrous convolution with gramian angular field representations is proposed in this paper. This approach breaks the barrier between one-dimensional vector and two-dimensional matrix by gramian angular field, transforming the spectral sequences from one-dimensional vectors into two-dimensional matrices, enabling the exploration of multidimensional relationships within spectral band relations through an atrous convolution-based spectral feature extraction network. The proposed model transcends the traditional one-dimensional spectral target detection limitations, offering a new perspective for spectral-based hyperspectral target detection. Experimental results on four real-world hyperspectral datasets demonstrate that the proposed method significantly outperforms existing state-of-the-art methods in detection performance, showcasing its potential for advancing hyperspectral target detection.

高光谱图像包含大量光谱波段，其丰富的光谱信息反映了物体的属性。事实证明，利用最先进的深度学习技术可以有效地进行高光谱目标检测。然而，与二维矩阵数据相比，光谱序列的一维性质限制了可提取的信息，给基于深度学习的高光谱目标检测方法带来了挑战。为解决这一问题，本文提出了一种新颖的高光谱目标检测方法，该方法采用了atrous卷积与gramian角场表示法。这种方法通过格兰角场打破了一维向量和二维矩阵之间的障碍，将光谱序列从一维向量转换为二维矩阵，从而能够通过基于无差卷积的光谱特征提取网络探索光谱波段关系中的多维关系。所提出的模型突破了传统一维光谱目标检测的限制，为基于光谱的高光谱目标检测提供了新的视角。在四个真实世界高光谱数据集上的实验结果表明，所提出的方法在检测性能上明显优于现有的最先进方法，展示了其在推进高光谱目标检测方面的潜力。

{"title":"Breaking dimensional barriers in hyperspectral target detection: Atrous convolution with Gramian Angular field representations","authors":"Hongzhou Wang , Yulei Wang , Yuchao Yang , Enyu Zhao , Jian Zeng","doi":"10.1016/j.infrared.2024.105623","DOIUrl":"10.1016/j.infrared.2024.105623","url":null,"abstract":"<div><div>Hyperspectral images contain extensive spectral bands with rich spectral information that reflects object properties. Leveraging state-of-the-art deep learning techniques has proven to be effective in hyperspectral target detection. However, compared to two-dimensional matrix data, the one-dimensional nature of spectral sequence limits the information that can be extracted, posing a challenge for deep learning-based hyperspectral target detection methodologies. To address this issue, a novel hyperspectral target detection method employing atrous convolution with gramian angular field representations is proposed in this paper. This approach breaks the barrier between one-dimensional vector and two-dimensional matrix by gramian angular field, transforming the spectral sequences from one-dimensional vectors into two-dimensional matrices, enabling the exploration of multidimensional relationships within spectral band relations through an atrous convolution-based spectral feature extraction network. The proposed model transcends the traditional one-dimensional spectral target detection limitations, offering a new perspective for spectral-based hyperspectral target detection. Experimental results on four real-world hyperspectral datasets demonstrate that the proposed method significantly outperforms existing state-of-the-art methods in detection performance, showcasing its potential for advancing hyperspectral target detection.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"143 ","pages":"Article 105623"},"PeriodicalIF":3.1,"publicationDate":"2024-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142655772","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

Multi-Scale convolutional neural network for finger vein recognition 用于手指静脉识别的多尺度卷积神经网络

IF 3.1 3区物理与天体物理 Q2 INSTRUMENTS & INSTRUMENTATION

Infrared Physics & Technology

Pub Date : 2024-11-12 DOI: 10.1016/j.infrared.2024.105624

Junbo Liu, Hui Ma, Zishuo Guo

With the continuous advancement of science and technology, an increasing number of deep learning methods are being applied in the field of finger vein recognition to describe the structural characteristics of finger veins. However, some deep learning methods fail to adequately extract longer texture features. during the feature extraction process, resulting in a decrease in the uniqueness of extracted finger vein features. Additionally, these methods tend to extract global information while neglecting the importance of local texture information. To address the aforementioned issues, this paper introduces a multiscale convolution network (MCNet) model based on finger vein structure. On one hand, a multiscale feature extraction (MFE) model based on the rectangular and square convolution kernels are employed to extract structural information from finger veins and to simultaneously enhance the features of longer texture features. On the other hand, the paper introduces a cross-information fusion attention (CFA) block that combines spatial and channel information, in order to enhance local details information and the network’s ability to extract vein patterns. The experimental results on the public datasets FV-USM, SDUMLA-HMT, and HKPU validate the effectiveness of MCNet with the recognition rates of 99.86%, 99.11%, and 99.15% respectively.

随着科学技术的不断进步，越来越多的深度学习方法被应用于手指静脉识别领域，以描述手指静脉的结构特征。然而，一些深度学习方法在特征提取过程中未能充分提取较长的纹理特征，导致提取的指静脉特征的唯一性降低。此外，这些方法倾向于提取全局信息，而忽视了局部纹理信息的重要性。针对上述问题，本文引入了基于指静脉结构的多尺度卷积网络（MCNet）模型。一方面，采用基于矩形和正方形卷积核的多尺度特征提取（MFE）模型来提取手指静脉的结构信息，并同时增强较长纹理特征的特性。另一方面，本文引入了交叉信息融合注意（CFA）模块，将空间信息和通道信息相结合，以增强局部细节信息和网络提取静脉模式的能力。在公共数据集 FV-USM、SDUMLA-HMT 和 HKPU 上的实验结果验证了 MCNet 的有效性，识别率分别为 99.86%、99.11% 和 99.15%。

{"title":"Multi-Scale convolutional neural network for finger vein recognition","authors":"Junbo Liu, Hui Ma, Zishuo Guo","doi":"10.1016/j.infrared.2024.105624","DOIUrl":"10.1016/j.infrared.2024.105624","url":null,"abstract":"<div><div>With the continuous advancement of science and technology, an increasing number of deep learning methods are being applied in the field of finger vein recognition to describe the structural characteristics of finger veins. However, some deep learning methods fail to adequately extract longer texture features. during the feature extraction process, resulting in a decrease in the uniqueness of extracted finger vein features. Additionally, these methods tend to extract global information while neglecting the importance of local texture information. To address the aforementioned issues, this paper introduces a multiscale convolution network (MCNet) model based on finger vein structure. On one hand, a multiscale feature extraction (MFE) model based on the rectangular and square convolution kernels are employed to extract structural information from finger veins and to simultaneously enhance the features of longer texture features. On the other hand, the paper introduces a cross-information fusion attention (CFA) block that combines spatial and channel information, in order to enhance local details information and the network’s ability to extract vein patterns. The experimental results on the public datasets FV-USM, SDUMLA-HMT, and HKPU validate the effectiveness of MCNet with the recognition rates of 99.86%, 99.11%, and 99.15% respectively.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"143 ","pages":"Article 105624"},"PeriodicalIF":3.1,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142655769","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

Temporal denoising and deep feature learning for enhanced defect detection in thermography using stacked denoising convolution autoencoder 使用堆叠去噪卷积自动编码器进行时间去噪和深度特征学习，以增强热成像中的缺陷检测能力

IF 3.1 3区物理与天体物理 Q2 INSTRUMENTS & INSTRUMENTATION

Infrared Physics & Technology

Pub Date : 2024-11-10 DOI: 10.1016/j.infrared.2024.105612

Naga Prasanthi Yerneni , V.S. Ghali , G.T. Vesala , Fei Wang , Ravibabu Mulaveesala

Thermal wave imaging uses the temporal temperature distribution over the object’s surface for subsurface analysis. However, the noise generated during experimentation corrupts this temporal history and hampers the detection of defect signatures. As denoising of the temporal thermal history enhances the defect detectability, this study offers a Stacked Denoising Convolution Autoencoder (SDCAE) in frequency-modulated thermal wave imaging with one-dimensional convolution layers to reduce noise in temporal thermal evolution and train high-level features resulting in improved defect signs. Experimental results on mild steel and carbon fiber reinforced polymer specimens with different sizes of defects at various depths demonstrate that integrating temporal denoising and deep feature learning techniques into a single novel framework significantly improved defect detectability. In addition, defect signal-to-noise ratios of the denoised thermal data and latent space of the proposed model compared to conventional autoencoder and dimensionality reduction techniques recommend the superiority of the proposed method.

热波成像利用物体表面的时间温度分布进行地下分析。然而，实验过程中产生的噪声会破坏这种时间历史，阻碍缺陷特征的检测。由于对时间热历史进行去噪可提高缺陷检测能力，本研究在频率调制热波成像中提供了叠加去噪卷积自动编码器（SDCAE），该编码器具有一维卷积层，可减少时间热演变中的噪声，并训练高级特征，从而改善缺陷特征。对不同深度不同尺寸缺陷的低碳钢和碳纤维增强聚合物试样的实验结果表明，将时间去噪和深度特征学习技术整合到一个新框架中，可显著提高缺陷的可探测性。此外，与传统的自动编码器和降维技术相比，拟议模型的去噪热数据和潜空间的缺陷信噪比证明了拟议方法的优越性。

{"title":"Temporal denoising and deep feature learning for enhanced defect detection in thermography using stacked denoising convolution autoencoder","authors":"Naga Prasanthi Yerneni , V.S. Ghali , G.T. Vesala , Fei Wang , Ravibabu Mulaveesala","doi":"10.1016/j.infrared.2024.105612","DOIUrl":"10.1016/j.infrared.2024.105612","url":null,"abstract":"<div><div>Thermal wave imaging uses the temporal temperature distribution over the object’s surface for subsurface analysis. However, the noise generated during experimentation corrupts this temporal history and hampers the detection of defect signatures. As denoising of the temporal thermal history enhances the defect detectability, this study offers a Stacked Denoising Convolution Autoencoder (SDCAE) in frequency-modulated thermal wave imaging with one-dimensional convolution layers to reduce noise in temporal thermal evolution and train high-level features resulting in improved defect signs. Experimental results on mild steel and carbon fiber reinforced polymer specimens with different sizes of defects at various depths demonstrate that integrating temporal denoising and deep feature learning techniques into a single novel framework significantly improved defect detectability. In addition, defect signal-to-noise ratios of the denoised thermal data and latent space of the proposed model compared to conventional autoencoder and dimensionality reduction techniques recommend the superiority of the proposed method.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"143 ","pages":"Article 105612"},"PeriodicalIF":3.1,"publicationDate":"2024-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142655768","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

Detection of black tea fermentation quality based on optimized deep neural network and hyperspectral imaging 基于优化深度神经网络和高光谱成像的红茶发酵质量检测

IF 3.1 3区物理与天体物理 Q2 INSTRUMENTS & INSTRUMENTATION

Infrared Physics & Technology

Pub Date : 2024-11-09 DOI: 10.1016/j.infrared.2024.105625

Minghao Huang , Yu Tang , Zhiping Tan , Jinchang Ren , Yong He , Huasheng Huang

The quality of black tea significantly relies on its fermentation process. Nevertheless, achieving precise and objective evaluations remains challenging due to the subjective nature of manual judgment involved in quality monitoring. To address this problem, hyperspectral imaging combined with the deep learning algorithms are proposed to identify the fermentation quality of black tea. Firstly, the hyperspectral data of Yinghong No. 9 black tea during five fermentation time intervals within 0–5 h are collected. Then, the Support Vector Machine (SVM), Artificial Neural Network (ANN), Partial Least Squares Discriminant Analysis (PLS-DA), and Naive Bayesian (NB) are used to construct black tea fermentation quality detection models based on full spectrum and selected spectrum data. Furthermore, deep learning algorithms including the Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), CNN-LSTM, and Swarm Optimization (PSO) optimized CNN-LSTM (PSO-CNN-LSTM) are also used to build the detection model using the spectral images. The experimental results indicate that deep learning algorithms have obvious advantages over traditional machine learning algorithms in tea fermentation quality detection. Besides, the PSO-CNN-LSTM model shows the best classification performance compared to other algorithms and achieves an accuracy of 96.78% on the test set. This study demonstrates the significant potential of combining deep learning with hyperspectral imaging for predicting black tea fermentation quality. This provides a new approach for effective monitoring of the black tea fermentation process and a useful reference for other applications in similar fields.

红茶的质量很大程度上取决于其发酵过程。然而，由于质量监测涉及人工判断的主观性，实现精确客观的评价仍具有挑战性。为解决这一问题，本文提出了高光谱成像结合深度学习算法来识别红茶的发酵质量。首先，采集英红九号红茶在 0-5 h 内五个发酵时间区间的高光谱数据。然后，利用支持向量机（SVM）、人工神经网络（ANN）、偏最小二乘判别分析（PLS-DA）和奈夫贝叶斯（NB）构建基于全谱和选谱数据的红茶发酵质量检测模型。此外，深度学习算法包括卷积神经网络（CNN）、长短期记忆（LSTM）、CNN-LSTM 和群优化（PSO）优化的 CNN-LSTM（PSO-CNN-LSTM），也被用于利用光谱图像建立检测模型。实验结果表明，与传统的机器学习算法相比，深度学习算法在茶叶发酵质量检测中具有明显的优势。此外，与其他算法相比，PSO-CNN-LSTM 模型的分类性能最好，在测试集上的准确率达到 96.78%。这项研究证明了深度学习与高光谱成像相结合在预测红茶发酵质量方面的巨大潜力。这为有效监测红茶发酵过程提供了一种新方法，也为类似领域的其他应用提供了有益的参考。

{"title":"Detection of black tea fermentation quality based on optimized deep neural network and hyperspectral imaging","authors":"Minghao Huang , Yu Tang , Zhiping Tan , Jinchang Ren , Yong He , Huasheng Huang","doi":"10.1016/j.infrared.2024.105625","DOIUrl":"10.1016/j.infrared.2024.105625","url":null,"abstract":"<div><div>The quality of black tea significantly relies on its fermentation process. Nevertheless, achieving precise and objective evaluations remains challenging due to the subjective nature of manual judgment involved in quality monitoring. To address this problem, hyperspectral imaging combined with the deep learning algorithms are proposed to identify the fermentation quality of black tea. Firstly, the hyperspectral data of Yinghong No. 9 black tea during five fermentation time intervals within 0–5 h are collected. Then, the Support Vector Machine (SVM), Artificial Neural Network (ANN), Partial Least Squares Discriminant Analysis (PLS-DA), and Naive Bayesian (NB) are used to construct black tea fermentation quality detection models based on full spectrum and selected spectrum data. Furthermore, deep learning algorithms including the Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), CNN-LSTM, and Swarm Optimization (PSO) optimized CNN-LSTM (PSO-CNN-LSTM) are also used to build the detection model using the spectral images. The experimental results indicate that deep learning algorithms have obvious advantages over traditional machine learning algorithms in tea fermentation quality detection. Besides, the PSO-CNN-LSTM model shows the best classification performance compared to other algorithms and achieves an accuracy of 96.78% on the test set. This study demonstrates the significant potential of combining deep learning with hyperspectral imaging for predicting black tea fermentation quality. This provides a new approach for effective monitoring of the black tea fermentation process and a useful reference for other applications in similar fields.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"143 ","pages":"Article 105625"},"PeriodicalIF":3.1,"publicationDate":"2024-11-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142655766","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

Hyperspectral and multispectral images fusion based on pyramid swin transformer 基于金字塔斯温变换器的高光谱和多光谱图像融合

IF 3.1 3区物理与天体物理 Q2 INSTRUMENTS & INSTRUMENTATION

Infrared Physics & Technology

Pub Date : 2024-11-07 DOI: 10.1016/j.infrared.2024.105617

Han Lang , Wenxing Bao , Wei Feng , Kewen Qu , Xuan Ma , Xiaowu Zhang

Remote sensing image fusion aims to generate a high spatial resolution hyperspectral image (HR-HSI) by integrating a low spatial resolution hyperspectral image (LR-HSI) and a high spatial resolution multispectral image (HR-MSI). While Convolutional Neural Networks (CNNs) have been widely employed in addressing the HSI-MSI fusion problem, their limited receptive field poses challenges in capturing global relationships within the feature maps. On the other hand, the computational complexity of Transformers hinders their application, especially in dealing with high-dimensional data like hyperspectral images (HSIs). To overcome this challenge, we propose an HSI-MSI fusion method based on the Pyramid Swin Transformer (PSTF). The pyramid design of the PSTF effectively extracts multi-scale information from images. The Spatial–Spectral Crossed Attention (SSCA) module, comprising the Cross Spatial Attention (CSA) and the Spectral Feature Integration (SFI) modules. The CSA module employs a cross-shaped self-attention mechanism, providing greater modeling flexibility for different spatial scales and non-local structures compared to traditional convolutional layers. Meanwhile, the SFI module introduces a global memory block (MB) to select the most relevant low-rank spectral vectors, integrating global spectral information with local spatial–spectral correlation to better extract and preserve spectral information. Additionally, the Separate Feature Extraction (SFE) module enhances the network’s ability to represent image features by independently processing positive and negative parts of shallow features, thus capturing details and structures more effectively and preventing the vanishing gradient problem. Compared with the state-of-the-art (SOTA) methods, experimental results demonstrate the effectiveness of the PSTF method.

遥感图像融合旨在通过整合低空间分辨率高光谱图像（LR-HSI）和高空间分辨率多光谱图像（HR-MSI），生成高空间分辨率高光谱图像（HR-HSI）。虽然卷积神经网络（CNN）已被广泛用于解决 HSI-MSI 融合问题，但其有限的感受野在捕捉特征图中的全局关系方面存在挑战。另一方面，变换器的计算复杂性也阻碍了其应用，尤其是在处理像高光谱图像（HSI）这样的高维数据时。为了克服这一难题，我们提出了一种基于金字塔斯温变换器（PSTF）的 HSI-MSI 融合方法。PSTF 的金字塔设计能有效提取图像中的多尺度信息。空间-光谱交叉注意（SSCA）模块由交叉空间注意（CSA）和光谱特征整合（SFI）模块组成。与传统的卷积层相比，CSA 模块采用了十字形自注意机制，为不同空间尺度和非局部结构提供了更大的建模灵活性。同时，SFI 模块引入了全局记忆块（MB），用于选择最相关的低秩光谱向量，将全局光谱信息与局部空间-光谱相关性整合在一起，从而更好地提取和保存光谱信息。此外，分离特征提取（SFE）模块通过独立处理浅层特征的正负部分，增强了网络表示图像特征的能力，从而更有效地捕捉细节和结构，防止梯度消失问题。与最先进的（SOTA）方法相比，实验结果证明了 PSTF 方法的有效性。

{"title":"Hyperspectral and multispectral images fusion based on pyramid swin transformer","authors":"Han Lang , Wenxing Bao , Wei Feng , Kewen Qu , Xuan Ma , Xiaowu Zhang","doi":"10.1016/j.infrared.2024.105617","DOIUrl":"10.1016/j.infrared.2024.105617","url":null,"abstract":"<div><div>Remote sensing image fusion aims to generate a high spatial resolution hyperspectral image (HR-HSI) by integrating a low spatial resolution hyperspectral image (LR-HSI) and a high spatial resolution multispectral image (HR-MSI). While Convolutional Neural Networks (CNNs) have been widely employed in addressing the HSI-MSI fusion problem, their limited receptive field poses challenges in capturing global relationships within the feature maps. On the other hand, the computational complexity of Transformers hinders their application, especially in dealing with high-dimensional data like hyperspectral images (HSIs). To overcome this challenge, we propose an HSI-MSI fusion method based on the Pyramid Swin Transformer (PSTF). The pyramid design of the PSTF effectively extracts multi-scale information from images. The Spatial–Spectral Crossed Attention (SSCA) module, comprising the Cross Spatial Attention (CSA) and the Spectral Feature Integration (SFI) modules. The CSA module employs a cross-shaped self-attention mechanism, providing greater modeling flexibility for different spatial scales and non-local structures compared to traditional convolutional layers. Meanwhile, the SFI module introduces a global memory block (MB) to select the most relevant low-rank spectral vectors, integrating global spectral information with local spatial–spectral correlation to better extract and preserve spectral information. Additionally, the Separate Feature Extraction (SFE) module enhances the network’s ability to represent image features by independently processing positive and negative parts of shallow features, thus capturing details and structures more effectively and preventing the vanishing gradient problem. Compared with the state-of-the-art (SOTA) methods, experimental results demonstrate the effectiveness of the PSTF method.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"143 ","pages":"Article 105617"},"PeriodicalIF":3.1,"publicationDate":"2024-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142655770","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

Small aircraft detection in infrared aerial imagery based on deep neural network 基于深度神经网络的红外航空图像中的小型飞机探测

IF 3.1 3区物理与天体物理 Q2 INSTRUMENTS & INSTRUMENTATION

Infrared Physics & Technology

Pub Date : 2024-11-04 DOI: 10.1016/j.infrared.2024.105454

Kai Zhang , Xiaotian Wang , Shaoyi Li , Bingyi Zhang

Detection of aerial target is an important part of infrared image processing. Both neural network method and traditional method can be used in infrared object detection. Neural network method has many advantages such as high accuracy and good portability compared with traditional object detection method. Since the features extracted by neural network method can change over detection target, automatic feature extraction makes neural network based detection method more effective. In recent years deep learning method has been also found wide use for object detection in images. In this paper, an object detection model based on the deep learning network YOLO is constructed for solving the infrared aircraft detection problem. We construct the dataset used for training and testing with recognized features being iteratively learned. The task of infrared object detection is sensitive to model size and detection speed. There is a requirement of using quantization method to reduce the storage space and the computation complexity. We propose a quantized model with appropriate accuracy for infrared object detection task. To solve the detection task for multiple extremely small aircrafts, model adjustment and quantization are used in proposed model and it gets a better performance. Experimental results on the constructed dataset show that the storage space for weight after quantization shrinks to a quarter, and there is no precision loss for extremely small aircrafts compared to the original model. The optimized YOLO-based deep learning model is effective to detect the small aircraft target in infrared aerial imagery.

空中目标检测是红外图像处理的重要组成部分。神经网络方法和传统方法都可用于红外目标检测。与传统的目标检测方法相比，神经网络方法具有精度高、便携性好等优点。由于神经网络方法提取的特征会随着检测目标的变化而变化，因此自动特征提取使得基于神经网络的检测方法更加有效。近年来，深度学习方法也被广泛应用于图像中的物体检测。本文构建了基于深度学习网络 YOLO 的物体检测模型，用于解决红外飞行器检测问题。我们构建了用于训练和测试的数据集，并通过迭代学习识别特征。红外物体检测任务对模型大小和检测速度非常敏感。需要使用量化方法来减少存储空间和计算复杂度。我们为红外物体检测任务提出了一个具有适当精度的量化模型。为了解决多架超小型飞机的检测任务，我们在所提出的模型中使用了模型调整和量化方法，并取得了较好的性能。在构建的数据集上的实验结果表明，量化后权重的存储空间缩小到四分之一，与原始模型相比，对极小飞机的检测没有精度损失。优化后的基于 YOLO 的深度学习模型可以有效地检测红外航空图像中的小型飞机目标。

{"title":"Small aircraft detection in infrared aerial imagery based on deep neural network","authors":"Kai Zhang , Xiaotian Wang , Shaoyi Li , Bingyi Zhang","doi":"10.1016/j.infrared.2024.105454","DOIUrl":"10.1016/j.infrared.2024.105454","url":null,"abstract":"<div><div>Detection of aerial target is an important part of infrared image processing. Both neural network method and traditional method can be used in infrared object detection. Neural network method has many advantages such as high accuracy and good portability compared with traditional object detection method. Since the features extracted by neural network method can change over detection target, automatic feature extraction makes neural network based detection method more effective. In recent years deep learning method has been also found wide use for object detection in images. In this paper, an object detection model based on the deep learning network YOLO is constructed for solving the infrared aircraft detection problem. We construct the dataset used for training and testing with recognized features being iteratively learned. The task of infrared object detection is sensitive to model size and detection speed. There is a requirement of using quantization method to reduce the storage space and the computation complexity. We propose a quantized model with appropriate accuracy for infrared object detection task. To solve the detection task for multiple extremely small aircrafts, model adjustment and quantization are used in proposed model and it gets a better performance. Experimental results on the constructed dataset show that the storage space for weight after quantization shrinks to a quarter, and there is no precision loss for extremely small aircrafts compared to the original model. The optimized YOLO-based deep learning model is effective to detect the small aircraft target in infrared aerial imagery.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"143 ","pages":"Article 105454"},"PeriodicalIF":3.1,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142655771","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

Improvement of mid-wavelength InAs/InAsSb nBn infrared detectors performance through interface control 通过接口控制提高中波长 InAs/InAsSb nBn 红外探测器的性能

IF 3.1 3区物理与天体物理 Q2 INSTRUMENTS & INSTRUMENTATION

Infrared Physics & Technology

Pub Date : 2024-11-03 DOI: 10.1016/j.infrared.2024.105619

Ye Zhang , Yifan Shan , Faran Chang , Yan Liang , Xiangyu Zhang , Guowei Wang , Donghai Wu , Dongwei Jiang , Hongyue Hao , Yingqiang Xu , Haiqiao Ni , Dan Lu , Zhichuan Niu

We report our study to optimize the growth of mid-wavelength InAs/InAsSb nBn infrared detectors through interface control method with AlSb/AlAs superlattices as electron barrier. The dark current model was employed to investigate the dominant dark current mechanism at various operating temperatures. We extracted the minority carrier lifetime of InAs/InAsSb material grown by different interface growth methods. Electrical and optical characterizations indicated superior performance of the device grown by migration-enhanced epitaxy (MEE) with a 3 s As and Sb soak time. With −0.3 V applied bias and 150 K operating temperature, the optimal device shown a dark current density of 8.95 × 10⁻⁶ A/cm² and peak specific detectivity of 7.12 × 10¹¹ cm Hz^1/2/W at 3.8 µm.

我们报告了以 AlSb/AlAs 超晶格为电子势垒，通过界面控制方法优化中波长 InAs/InAsSb nBn 红外探测器生长的研究。我们采用暗电流模型研究了不同工作温度下的主导暗电流机制。我们提取了采用不同界面生长方法生长的 InAs/InAsSb 材料的少数载流子寿命。电学和光学特性分析表明，通过迁移增强外延（MEE）生长的器件在 3 秒砷和锑浸泡时间内具有卓越的性能。在 -0.3 V 外加偏压和 150 K 工作温度下，最佳器件的暗电流密度为 8.95 × 10-6 A/cm2，在 3.8 µm 处的峰值比检测率为 7.12 × 1011 cm Hz1/2/W。

引用次数: 0

Optimization and effect comparison of typical gas pressure compensation model in chemical industry park 化工园区典型气体压力补偿模型的优化与效果比较

IF 3.1 3区物理与天体物理 Q2 INSTRUMENTS & INSTRUMENTATION

Infrared Physics & Technology

Pub Date : 2024-11-03 DOI: 10.1016/j.infrared.2024.105621

Fuchao Tian , Xinyu Xiang , Lejing Qin , Jiliang Huang , Bo Tan

In chemical parks, the leakage of harmful gases can lead to poisoning and even explosions. Therefore, its monitoring and leakage warnings are crucial. This paper takes the harmful gases CO₂, CH₄, and CO as examples, and set up an infrared gas sensor pressure compensation experimental platform to carry out experiments, to solve the problem of decreased detection accuracy of gas sensors due to changes in ambient pressure during detection. The pressure compensation experimental ranges of the gas sensors are 0 ∼ 5 %, 0 ∼ 20 %, and 0 ∼ 1000 ppm, and the maximum absolute errors of the infrared gas test data obtained under different concentrations and pressures are 0.24 ∼ 0.67, 0.89 ∼ 1.12, and 45 ∼ 60 ppm, respectively. The pressure compensation model based on the least squares method was constructed, and the maximum absolute errors were obtained as 0.08 ∼ 0.19, 0.13 ∼ 0.64, and 24 ∼ 37 ppm, respectively. The pressure compensation model based on the GA-BP neural network was constructed, and the maximum absolute errors were 0.04 ∼ 0.10, 0.08 ∼ 0.10, and 0.60 ∼ 8.30 ppm, respectively. The GA-BP neural network combines the genetic algorithm and the backpropagation algorithm, which can better deal with nonlinear problems. The comparison of these two models reflects the superiority of the GA-BP neural network model in the compensation effect. The establishment of the neural network pressure compensation model optimized by the genetic algorithm can effectively improve the detection accuracy of the gas sensor, and it is expected that the results are of great practical significance to guarantee production safety and protect the environment in the enterprise chemical park.

在化工园区，有害气体的泄漏会导致中毒甚至爆炸。因此，对其进行监测和泄漏预警至关重要。本文以有害气体 CO2、CH4、CO 为例，建立红外气体传感器压力补偿实验平台进行实验，以解决气体传感器在检测过程中因环境压力变化而导致检测精度下降的问题。气体传感器的压力补偿实验范围分别为 0 ∼ 5 %、0 ∼ 20 % 和 0 ∼ 1000 ppm，在不同浓度和压力下获得的红外气体检测数据的最大绝对误差分别为 0.24 ∼ 0.67、0.89 ∼ 1.12 和 45 ∼ 60 ppm。构建了基于最小二乘法的压力补偿模型，得到的最大绝对误差分别为 0.08 ∼ 0.19、0.13 ∼ 0.64 和 24 ∼ 37 ppm。构建了基于 GA-BP 神经网络的压力补偿模型，其最大绝对误差分别为 0.04 ∼ 0.10、0.08 ∼ 0.10 和 0.60 ∼ 8.30 ppm。GA-BP 神经网络结合了遗传算法和反向传播算法，能更好地处理非线性问题。这两种模型的比较反映了 GA-BP 神经网络模型在补偿效果上的优越性。通过遗传算法优化的神经网络压力补偿模型的建立，可有效提高气体传感器的检测精度，预计其结果对保障企业化工园区的生产安全和环境保护具有重要的现实意义。

{"title":"Optimization and effect comparison of typical gas pressure compensation model in chemical industry park","authors":"Fuchao Tian , Xinyu Xiang , Lejing Qin , Jiliang Huang , Bo Tan","doi":"10.1016/j.infrared.2024.105621","DOIUrl":"10.1016/j.infrared.2024.105621","url":null,"abstract":"<div><div>In chemical parks, the leakage of harmful gases can lead to poisoning and even explosions. Therefore, its monitoring and leakage warnings are crucial. This paper takes the harmful gases CO<sub>2</sub>, CH<sub>4</sub>, and CO as examples, and set up an infrared gas sensor pressure compensation experimental platform to carry out experiments, to solve the problem of decreased detection accuracy of gas sensors due to changes in ambient pressure during detection. The pressure compensation experimental ranges of the gas sensors are 0 ∼ 5 %, 0 ∼ 20 %, and 0 ∼ 1000 ppm, and the maximum absolute errors of the infrared gas test data obtained under different concentrations and pressures are 0.24 ∼ 0.67, 0.89 ∼ 1.12, and 45 ∼ 60 ppm, respectively. The pressure compensation model based on the least squares method was constructed, and the maximum absolute errors were obtained as 0.08 ∼ 0.19, 0.13 ∼ 0.64, and 24 ∼ 37 ppm, respectively. The pressure compensation model based on the GA-BP neural network was constructed, and the maximum absolute errors were 0.04 ∼ 0.10, 0.08 ∼ 0.10, and 0.60 ∼ 8.30 ppm, respectively. The GA-BP neural network combines the genetic algorithm and the backpropagation algorithm, which can better deal with nonlinear problems. The comparison of these two models reflects the superiority of the GA-BP neural network model in the compensation effect. The establishment of the neural network pressure compensation model optimized by the genetic algorithm can effectively improve the detection accuracy of the gas sensor, and it is expected that the results are of great practical significance to guarantee production safety and protect the environment in the enterprise chemical park.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"143 ","pages":"Article 105621"},"PeriodicalIF":3.1,"publicationDate":"2024-11-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142655763","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

K-means adaptive 2DSSA based on sparse representation model for hyperspectral target detection 基于稀疏表示模型的 K-means 自适应 2DSSA 用于高光谱目标检测

IF 3.1 3区物理与天体物理 Q2 INSTRUMENTS & INSTRUMENTATION

Infrared Physics & Technology

Pub Date : 2024-10-30 DOI: 10.1016/j.infrared.2024.105616

Tianshu Zhou , Yi Cen , Jiani He , Yueming Wang

Target detection is a hot spot in hyperspectral imagery (HSI) processing. The detection accuracy of target detection algorithms based on sparse representation (SR) models usually suffers from the high reconstruction residuals caused by inaccurate background estimations and insufficient target samples. Besides, with the development of hyperspectral imaging technology, the spatial resolution of HSI has been continuously enhanced, which can provide more spatial information for target detection. However, spatial information is often overlooked, leading to the underutilization of the pluralistic features of HSI. Target detection using only spectral information is susceptible to spectral variation, resulting in a high false alarm rate. To alleviate these problems, this paper proposes a joint spatial-spectral algorithm. In terms of spectra, a dictionary construction strategy (DCS) is designed for the sparse representation-based binary hypothesis (SRBBH) detector to reduce reconstruction residuals of target and background samples. In terms of space, k-means 2D adaptive singular spectrum analysis (KSSA) is used to extract spatial features in cluster units. Using spatial features can enhance the robustness of the algorithm to spectral variation, thereby reducing false alarms. The target detection results are obtained by applying DCS-SRBBH to the KSSA feature image. We evaluate the proposed algorithm on three datasets: two public and one of our own. Comprehensive experimental results indicate that the proposed algorithm outperforms other target detection algorithms in terms of accuracy.

目标检测是高光谱成像（HSI）处理中的一个热点。基于稀疏表示（SR）模型的目标检测算法的检测精度通常受到背景估计不准确和目标样本不足导致的重建残差过高的影响。此外，随着高光谱成像技术的发展，HSI 的空间分辨率不断提高，可以为目标检测提供更多的空间信息。然而，空间信息往往被忽视，导致高光谱成像的多元特征未得到充分利用。仅使用光谱信息进行目标检测容易受到光谱变化的影响，导致误报率较高。为了缓解这些问题，本文提出了一种空间-光谱联合算法。在光谱方面，为基于稀疏表示的二元假设（SRBBH）检测器设计了字典构建策略（DCS），以减少目标和背景样本的重建残差。在空间方面，K-means 二维自适应奇异谱分析（KSSA）用于提取聚类单元中的空间特征。使用空间特征可以增强算法对频谱变化的鲁棒性，从而减少误报。将 DCS-SRBBH 应用于 KSSA 特征图像可获得目标检测结果。我们在三个数据集上对所提出的算法进行了评估：两个公共数据集和一个我们自己的数据集。综合实验结果表明，所提出的算法在准确性方面优于其他目标检测算法。

{"title":"K-means adaptive 2DSSA based on sparse representation model for hyperspectral target detection","authors":"Tianshu Zhou , Yi Cen , Jiani He , Yueming Wang","doi":"10.1016/j.infrared.2024.105616","DOIUrl":"10.1016/j.infrared.2024.105616","url":null,"abstract":"<div><div>Target detection is a hot spot in hyperspectral imagery (HSI) processing. The detection accuracy of target detection algorithms based on sparse representation (SR) models usually suffers from the high reconstruction residuals caused by inaccurate background estimations and insufficient target samples. Besides, with the development of hyperspectral imaging technology, the spatial resolution of HSI has been continuously enhanced, which can provide more spatial information for target detection. However, spatial information is often overlooked, leading to the underutilization of the pluralistic features of HSI. Target detection using only spectral information is susceptible to spectral variation, resulting in a high false alarm rate. To alleviate these problems, this paper proposes a joint spatial-spectral algorithm. In terms of spectra, a dictionary construction strategy (DCS) is designed for the sparse representation-based binary hypothesis (SRBBH) detector to reduce reconstruction residuals of target and background samples. In terms of space, k-means 2D adaptive singular spectrum analysis (KSSA) is used to extract spatial features in cluster units. Using spatial features can enhance the robustness of the algorithm to spectral variation, thereby reducing false alarms. The target detection results are obtained by applying DCS-SRBBH to the KSSA feature image. We evaluate the proposed algorithm on three datasets: two public and one of our own. Comprehensive experimental results indicate that the proposed algorithm outperforms other target detection algorithms in terms of accuracy.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"143 ","pages":"Article 105616"},"PeriodicalIF":3.1,"publicationDate":"2024-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142655764","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 miniaturized multi-mechanism resonance-enhanced fiber optic photoacoustic multi-gas sensor 微型多机制共振增强光纤光声多气体传感器

IF 3.1 3区物理与天体物理 Q2 INSTRUMENTS & INSTRUMENTATION

Infrared Physics & Technology

Pub Date : 2024-10-29 DOI: 10.1016/j.infrared.2024.105615

Guojie Wu , Yuchen Guan , Zhenfeng Gong , Xue Wu , Liang Mei

A multi-mechanism resonance-enhanced fiber-optic photoacoustic multi-gas sensor (MR-FOPMS) is reported, in which the acoustic resonance of the T-type resonator is employed for C₂H₂ gas sensing and the mechanical resonance of the silicon cantilever is employed for the detection of CH₄ gas sensing. The silicon cantilever fiber-optic microphone and T-type photoacoustic resonator are designed using theoretical and finite element methods. The optimized cavity volume of the entire sensor is only 9.3 cm³ and the optimized silicon cantilever microphone has an ultra-high sensitivity of 62540.2 nm/Pa at the resonance. The ability of the sensor to detect multiple gases is demonstrated by simultaneous measurement of C₂H₂ and CH₄ using two DFB lasers at 1532.8 nm and 1650.96 nm as excitation sources. The lowest detection limits of the sensor are determined to be 158 and 382 ppb for C₂H₂ and CH₄, respectively, corresponding to normalized noise equivalent absorption coefficients of 2.82 × 10^-9, 1.43 × 10^-9 cm⁻¹ W Hz^−1/2, respectively.

报告了一种多机制共振增强光纤光声多气体传感器（MR-FOPMS），其中利用 T 型谐振器的声学共振进行 C2H2 气体传感，利用硅悬臂的机械共振进行 CH4 气体传感检测。采用理论和有限元方法设计了硅悬臂光纤传声器和 T 型光声谐振器。整个传感器的优化腔体体积仅为 9.3 cm3，优化后的硅悬臂传声器在共振时具有 62540.2 nm/Pa 的超高灵敏度。使用 1532.8 nm 和 1650.96 nm 的两个 DFB 激光作为激发光源同时测量 C2H2 和 CH4，证明了该传感器检测多种气体的能力。传感器对 C2H2 和 CH4 的最低检测限分别为 158 和 382 ppb，对应的归一化噪声等效吸收系数分别为 2.82 × 10-9 和 1.43 × 10-9 cm-1 W Hz-1/2。

{"title":"A miniaturized multi-mechanism resonance-enhanced fiber optic photoacoustic multi-gas sensor","authors":"Guojie Wu , Yuchen Guan , Zhenfeng Gong , Xue Wu , Liang Mei","doi":"10.1016/j.infrared.2024.105615","DOIUrl":"10.1016/j.infrared.2024.105615","url":null,"abstract":"<div><div>A multi-mechanism resonance-enhanced fiber-optic photoacoustic multi-gas sensor (MR-FOPMS) is reported, in which the acoustic resonance of the T-type resonator is employed for C<sub>2</sub>H<sub>2</sub> gas sensing and the mechanical resonance of the silicon cantilever is employed for the detection of CH<sub>4</sub> gas sensing. The silicon cantilever fiber-optic microphone and T-type photoacoustic resonator are designed using theoretical and finite element methods. The optimized cavity volume of the entire sensor is only 9.3 cm<sup>3</sup> and the optimized silicon cantilever microphone has an ultra-high sensitivity of 62540.2 nm/Pa at the resonance. The ability of the sensor to detect multiple gases is demonstrated by simultaneous measurement of C<sub>2</sub>H<sub>2</sub> and CH<sub>4</sub> using two DFB lasers at 1532.8 nm and 1650.96 nm as excitation sources. The lowest detection limits of the sensor are determined to be 158 and 382 ppb for C<sub>2</sub>H<sub>2</sub> and CH<sub>4</sub>, respectively, corresponding to normalized noise equivalent absorption coefficients of 2.82 × 10<sup>-9</sup>, 1.43 × 10<sup>-9</sup> cm<sup>−1</sup> W Hz<sup>−1/2</sup>, respectively.</div></div>","PeriodicalId":13549,"journal":{"name":"Infrared Physics & Technology","volume":"143 ","pages":"Article 105615"},"PeriodicalIF":3.1,"publicationDate":"2024-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142655765","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