Pub Date : 2025-06-26DOI: 10.1109/TCI.2025.3583465
Zhuang He;Hai-Miao Hu;Likun Gao;Haoxin Hu;Xinhui Xue;Zhenglin Tang;Difeng Zhu;Haowen Zheng;Chongze Wang
Vignetting correction is an essential process of image signal processing. It is an important part for obtaining high-quality images, but the research in this field has not been fully emphasized. The mainstream methods are based on calibration which processes are complex. And many methods get low accuracy and poor robustness in practical. In this paper, we analyzed the optical principle of vignetting and its influence on the image. Then, we proposed an algorithm based on color-intensity map entropy optimization to correct image vignetting. Moreover, because of the lack of dataset of vignetting, we proposed a method for constructing vignetting image dataset through capturing the real scenes. Compared with the dataset generated through simulation, our dataset is more authentic and reliable. Many experiments have been carried out on this dataset, and the results proved that the proposed algorithm achieved the best performance.
{"title":"Vignetting Correction Through Color-Intensity Map Entropy Optimization","authors":"Zhuang He;Hai-Miao Hu;Likun Gao;Haoxin Hu;Xinhui Xue;Zhenglin Tang;Difeng Zhu;Haowen Zheng;Chongze Wang","doi":"10.1109/TCI.2025.3583465","DOIUrl":"https://doi.org/10.1109/TCI.2025.3583465","url":null,"abstract":"Vignetting correction is an essential process of image signal processing. It is an important part for obtaining high-quality images, but the research in this field has not been fully emphasized. The mainstream methods are based on calibration which processes are complex. And many methods get low accuracy and poor robustness in practical. In this paper, we analyzed the optical principle of vignetting and its influence on the image. Then, we proposed an algorithm based on color-intensity map entropy optimization to correct image vignetting. Moreover, because of the lack of dataset of vignetting, we proposed a method for constructing vignetting image dataset through capturing the real scenes. Compared with the dataset generated through simulation, our dataset is more authentic and reliable. Many experiments have been carried out on this dataset, and the results proved that the proposed algorithm achieved the best performance.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"11 ","pages":"911-925"},"PeriodicalIF":4.2,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144623935","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-06-19DOI: 10.1109/TCI.2025.3581105
Jie Tang;Gaofeng Peng;Jialu Liu;Bo Yu
Developing fast and accurate stereo matching algorithms is crucial for real-world embedded vision applications. Depth information plays a significant role in scene understanding, and depth calculated through stereo matching is generally considered to be more precise and reliable than that obtained from monocular depth estimation. However, speed-oriented stereo matching methods often suffer from poor feature representation due to sparse sampling and detail loss caused by unreasonable disparity allocation during upsampling. To address these issues, we propose G2L-Stereo, a two-stage real-time stereo matching network that combines global disparity range prediction and local disparity range prediction. In the global disparity range prediction stage, we introduce feature-guided connections for cost aggregation, enhancing the expressive power of sparse features by aligning the feature space across different scales of cost volumes. We also incorporate confidence estimation into the upsampling algorithm to reduce the propagation of inaccurate disparities during upsampling, yielding more precise disparity maps. In the local disparity range prediction stage, we develop a disparity refinement module guided by neighborhood similarity. This module aggregates similar neighboring costs to estimate disparity residuals and refine disparities, restoring lost details in the low-resolution disparity map and further enhancing disparity accuracy. Extensive experiments on the SceneFlow and KITTI datasets validate the effectiveness of our model, showing that G2L-Stereo achieves fast inference while maintaining accuracy comparable to state-of-the-art methods.
{"title":"G2L-Stereo: Global to Local Two-Stage Real-Time Stereo Matching Network","authors":"Jie Tang;Gaofeng Peng;Jialu Liu;Bo Yu","doi":"10.1109/TCI.2025.3581105","DOIUrl":"https://doi.org/10.1109/TCI.2025.3581105","url":null,"abstract":"Developing fast and accurate stereo matching algorithms is crucial for real-world embedded vision applications. Depth information plays a significant role in scene understanding, and depth calculated through stereo matching is generally considered to be more precise and reliable than that obtained from monocular depth estimation. However, speed-oriented stereo matching methods often suffer from poor feature representation due to sparse sampling and detail loss caused by unreasonable disparity allocation during upsampling. To address these issues, we propose G2L-Stereo, a two-stage real-time stereo matching network that combines global disparity range prediction and local disparity range prediction. In the global disparity range prediction stage, we introduce feature-guided connections for cost aggregation, enhancing the expressive power of sparse features by aligning the feature space across different scales of cost volumes. We also incorporate confidence estimation into the upsampling algorithm to reduce the propagation of inaccurate disparities during upsampling, yielding more precise disparity maps. In the local disparity range prediction stage, we develop a disparity refinement module guided by neighborhood similarity. This module aggregates similar neighboring costs to estimate disparity residuals and refine disparities, restoring lost details in the low-resolution disparity map and further enhancing disparity accuracy. Extensive experiments on the SceneFlow and KITTI datasets validate the effectiveness of our model, showing that G2L-Stereo achieves fast inference while maintaining accuracy comparable to state-of-the-art methods.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"11 ","pages":"852-863"},"PeriodicalIF":4.2,"publicationDate":"2025-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144524418","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Spectral CT can be used to perform material decomposition from polychromatic attenuation data, generate virtual monochromatic or virtual narrow-energy-width images in which beam hardening artifacts are suppressed, and provide detailed energy attenuation coefficients for material characterization. We propose an energy-coded spectral CT imaging method that is based on projection mix separation, which enables simultaneous energy decoding and image reconstruction. An X-ray energy-coded forward model is then constructed. Leveraging the Poisson statistical properties of the measurement data, we formulate a constrained optimization problem for both the energy-coded coefficient matrix and the material decomposition coefficient matrix, which is solved using a block coordinate descent algorithm. Simulations and experimental results demonstrate that the decoded energy spectrum distribution and virtual narrow-energy-width CT images are accurate and effective. The proposed method suppresses beam hardening artifacts and enhances the material identification capabilities of traditional CT.
{"title":"Energy-Coded Spectral CT Imaging Method Based on Projection Mix Separation","authors":"Xiaojie Zhao;Yihong Li;Yan Han;Ping Chen;Jiaotong Wei","doi":"10.1109/TCI.2025.3578762","DOIUrl":"https://doi.org/10.1109/TCI.2025.3578762","url":null,"abstract":"Spectral CT can be used to perform material decomposition from polychromatic attenuation data, generate virtual monochromatic or virtual narrow-energy-width images in which beam hardening artifacts are suppressed, and provide detailed energy attenuation coefficients for material characterization. We propose an energy-coded spectral CT imaging method that is based on projection mix separation, which enables simultaneous energy decoding and image reconstruction. An X-ray energy-coded forward model is then constructed. Leveraging the Poisson statistical properties of the measurement data, we formulate a constrained optimization problem for both the energy-coded coefficient matrix and the material decomposition coefficient matrix, which is solved using a block coordinate descent algorithm. Simulations and experimental results demonstrate that the decoded energy spectrum distribution and virtual narrow-energy-width CT images are accurate and effective. The proposed method suppresses beam hardening artifacts and enhances the material identification capabilities of traditional CT.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"11 ","pages":"839-851"},"PeriodicalIF":4.2,"publicationDate":"2025-06-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144492200","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
High-resolution medical images can provide more detailed information for better diagnosis. Conventional medical image super-resolution relies on a single task which first performs the extraction of the features and then upscaling based on the features. The features extracted may not be complete for super-resolution. Recent multi-task learning, including reconstruction and super-resolution, is a good solution to obtain additional relevant information. The interaction between the two tasks is often insufficient, which still leads to incomplete and less relevant deep features. To address above limitations, we propose an iterative collaboration network (ICONet) to improve communications between tasks by progressively incorporating reconstruction prior to the super-resolution learning procedure in an iterative collaboration way. It consists of a reconstruction branch, a super-resolution branch, and a SR-Rec fusion module. The reconstruction branch generates the artifact-free image as prior, which is followed by a super-resolution branch for prior knowledge-guided super-resolution. Unlike the widely-used convolutional neural networks for extracting local features and Transformers with quadratic computational complexity for modeling long-range dependencies, we develop a new residual spatial-channel feature learning (RSCFL) module of two branches to efficiently establish feature relationships in spatial and channel dimensions. Moreover, the designed SR-Rec fusion module fuses the reconstruction prior and super-resolution features with each other in an adaptive manner. Our ICONet is built with multi-stage models to iteratively upscale the low-resolution images using steps of ${2 times }$ and simultaneously interact between two branches in multi-stage supervisions. Quantitative and qualitative experimental results on the benchmarking dataset show that our ICONet outperforms most state-of-the-art approaches.
{"title":"Iterative Collaboration Network Guided by Reconstruction Prior for Medical Image Super-Resolution","authors":"Xiaoyan Kui;Zexin Ji;Beiji Zou;Yang Li;Yulan Dai;Liming Chen;Pierre Vera;Su Ruan","doi":"10.1109/TCI.2025.3577340","DOIUrl":"https://doi.org/10.1109/TCI.2025.3577340","url":null,"abstract":"High-resolution medical images can provide more detailed information for better diagnosis. Conventional medical image super-resolution relies on a single task which first performs the extraction of the features and then upscaling based on the features. The features extracted may not be complete for super-resolution. Recent multi-task learning, including reconstruction and super-resolution, is a good solution to obtain additional relevant information. The interaction between the two tasks is often insufficient, which still leads to incomplete and less relevant deep features. To address above limitations, we propose an iterative collaboration network (ICONet) to improve communications between tasks by progressively incorporating reconstruction prior to the super-resolution learning procedure in an iterative collaboration way. It consists of a reconstruction branch, a super-resolution branch, and a SR-Rec fusion module. The reconstruction branch generates the artifact-free image as prior, which is followed by a super-resolution branch for prior knowledge-guided super-resolution. Unlike the widely-used convolutional neural networks for extracting local features and Transformers with quadratic computational complexity for modeling long-range dependencies, we develop a new residual spatial-channel feature learning (RSCFL) module of two branches to efficiently establish feature relationships in spatial and channel dimensions. Moreover, the designed SR-Rec fusion module fuses the reconstruction prior and super-resolution features with each other in an adaptive manner. Our ICONet is built with multi-stage models to iteratively upscale the low-resolution images using steps of <inline-formula> <tex-math>${2 times }$</tex-math></inline-formula> and simultaneously interact between two branches in multi-stage supervisions. Quantitative and qualitative experimental results on the benchmarking dataset show that our ICONet outperforms most state-of-the-art approaches.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"11 ","pages":"827-838"},"PeriodicalIF":4.2,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144336075","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Transformers bring significantly improved performance to the light field image super-resolution task due to their long-range dependency modeling capability. However, the inherently high computational complexity of their core self-attention mechanism has increasingly hindered their advancement in this task. To address this issue, we first introduce the LF-VSSM block, a novel module inspired by progressive feature extraction, to efficiently capture critical long-range spatial-angular dependencies in light field images. LF-VSSM successively extracts spatial features within sub-aperture images, spatial-angular features between sub-aperture images, and spatial-angular features between light field image pixels. On this basis, we propose a lightweight network, $L^{2}$FMamba (Lightweight Light Field Mamba), which integrates the LF-VSSM block to leverage light field features for super-resolution tasks while overcoming the computational challenges of Transformer-based approaches. Extensive experiments on multiple light field datasets demonstrate that our method reduces the number of parameters and complexity while achieving superior super-resolution performance with faster inference speed.
{"title":"$L^{2}$FMamba: Lightweight Light Field Image Super-Resolution With State Space Model","authors":"Zeqiang Wei;Kai Jin;Zeyi Hou;Kuan Song;Xiuzhuang Zhou","doi":"10.1109/TCI.2025.3577338","DOIUrl":"https://doi.org/10.1109/TCI.2025.3577338","url":null,"abstract":"Transformers bring significantly improved performance to the light field image super-resolution task due to their long-range dependency modeling capability. However, the inherently high computational complexity of their core self-attention mechanism has increasingly hindered their advancement in this task. To address this issue, we first introduce the LF-VSSM block, a novel module inspired by progressive feature extraction, to efficiently capture critical long-range spatial-angular dependencies in light field images. LF-VSSM successively extracts spatial features within sub-aperture images, spatial-angular features between sub-aperture images, and spatial-angular features between light field image pixels. On this basis, we propose a lightweight network, <inline-formula><tex-math>$L^{2}$</tex-math></inline-formula>FMamba (Lightweight Light Field Mamba), which integrates the LF-VSSM block to leverage light field features for super-resolution tasks while overcoming the computational challenges of Transformer-based approaches. Extensive experiments on multiple light field datasets demonstrate that our method reduces the number of parameters and complexity while achieving superior super-resolution performance with faster inference speed.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"11 ","pages":"816-826"},"PeriodicalIF":4.2,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144323022","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-06-06DOI: 10.1109/TCI.2025.3577405
Jian Song;Fatemeh Pourahmadian;Todd W. Murray;Venkatalakshmi V. Narumanchi
This study investigates the imaging ability of the time-domain linear sampling method (TLSM) when applied to laser ultrasonic (LU) tomography of subsurface defects from limited-aperture measurements. In this vein, the TLSM indicator and its spectral counterpart known as the multifrequency LSM are formulated within the context of LU testing. The affiliated imaging functionals are then computed using synthetic and experimental data germane to LU inspection of aluminum alloy specimens with manufactured defects. Hyperparameters of inversion are computationally analyzed. We demonstrate using synthetic data that the TLSM indicator has the unique ability to recover weak (or hard-to-reach) scatterers and has the potential to generate higher quality images compared to LSM. Provided high-SNR measurements, this advantage may be preserved in reconstructions from LU test data.
{"title":"Laser Ultrasonic Imaging Via the Time Domain Linear Sampling Method","authors":"Jian Song;Fatemeh Pourahmadian;Todd W. Murray;Venkatalakshmi V. Narumanchi","doi":"10.1109/TCI.2025.3577405","DOIUrl":"https://doi.org/10.1109/TCI.2025.3577405","url":null,"abstract":"This study investigates the imaging ability of the time-domain linear sampling method (TLSM) when applied to laser ultrasonic (LU) tomography of subsurface defects from limited-aperture measurements. In this vein, the TLSM indicator and its spectral counterpart known as the multifrequency LSM are formulated within the context of LU testing. The affiliated imaging functionals are then computed using synthetic and experimental data germane to LU inspection of aluminum alloy specimens with manufactured defects. Hyperparameters of inversion are computationally analyzed. We demonstrate using synthetic data that the TLSM indicator has the unique ability to recover weak (or hard-to-reach) scatterers and has the potential to generate higher quality images compared to LSM. Provided high-SNR measurements, this advantage may be preserved in reconstructions from LU test data.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"11 ","pages":"803-815"},"PeriodicalIF":4.2,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144323023","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-06-06DOI: 10.1109/TCI.2025.3577334
Zhihan Xu;Yin Xiao;Wen Chen
Random disturbance has become a great challenge for correspondence imaging (CI) due to dynamic and nonlinear scaling factors. In this paper, we propose a robust CI against random disturbances for high-quality object reconstruction. To remove the effect of dynamic scaling factors induced by random disturbance, a wavelet and total variation (WATV) algorithm is developed to estimate a series of varying thresholds. Then, light intensities collected by a single-pixel detector are processed by using the series of estimated varying thresholds. To realize high-quality object reconstruction, the binarized light intensities and a series of random patterns are fed into a plug-and-play priors (PnP) algorithm with an iteration framework and a general denoiser, called as CI-PnP. Theoretical descriptions are given in detail to reveal the formation mechanism in CI under random disturbance. Optical measurements are conducted to verify robustness of the proposed CI against random disturbances. It is demonstrated that the proposed method can remove the effect of dynamic scaling factors induced by random disturbance, and can realize high-quality object reconstruction. The proposed method provides a promising solution to achieving ultra-high robustness against random disturbances in CI, and is promising in various applications.
{"title":"Robust Correspondence Imaging Against Random Disturbances With Single-Pixel Detection","authors":"Zhihan Xu;Yin Xiao;Wen Chen","doi":"10.1109/TCI.2025.3577334","DOIUrl":"https://doi.org/10.1109/TCI.2025.3577334","url":null,"abstract":"Random disturbance has become a great challenge for correspondence imaging (CI) due to dynamic and nonlinear scaling factors. In this paper, we propose a robust CI against random disturbances for high-quality object reconstruction. To remove the effect of dynamic scaling factors induced by random disturbance, a wavelet and total variation (WATV) algorithm is developed to estimate a series of varying thresholds. Then, light intensities collected by a single-pixel detector are processed by using the series of estimated varying thresholds. To realize high-quality object reconstruction, the binarized light intensities and a series of random patterns are fed into a plug-and-play priors (PnP) algorithm with an iteration framework and a general denoiser, called as CI-PnP. Theoretical descriptions are given in detail to reveal the formation mechanism in CI under random disturbance. Optical measurements are conducted to verify robustness of the proposed CI against random disturbances. It is demonstrated that the proposed method can remove the effect of dynamic scaling factors induced by random disturbance, and can realize high-quality object reconstruction. The proposed method provides a promising solution to achieving ultra-high robustness against random disturbances in CI, and is promising in various applications.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"11 ","pages":"901-910"},"PeriodicalIF":4.2,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144581738","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-06-06DOI: 10.1109/TCI.2025.3572250
Han Yue;Jun Cheng;Yu-Xuan Ren;Chien-Chun Chen;Grant A. van Riessen;Philip Heng Wai Leong;Steve Feng Shu
Ptychographic imaging confronts inherent challenges in applying deep learning for phase retrieval from diffraction patterns. Conventional neural architectures, both convolutional neural networks and Transformer-based methods, are optimized for natural images with Euclidean spatial neighborhood-based inductive biases that exhibit geometric mismatch with the concentric coherent patterns characteristic of diffraction data in reciprocal space. In this paper, we present PPN, a physics-inspired deep learning network with Polar Coordinate Attention (PoCA) for ptychographic imaging, that aligns neural inductive biases with diffraction physics through a dual-branch architecture separating local feature extraction from non-local coherence modeling. It consists of a PoCA mechanism that replaces Euclidean spatial priors with physically consistent radial-angular correlations. PPN outperforms existing end-to-end models, with spectral and spatial analysis confirming its greater preservation of high-frequency details. Notably, PPN maintains robust performance compared to iterative methods even at low overlap ratios — well-suited for high-throughput imaging in real-world acquisition scenarios for samples with consistent structural characteristics.
{"title":"A Physics-Inspired Deep Learning Framework With Polar Coordinate Attention for Ptychographic Imaging","authors":"Han Yue;Jun Cheng;Yu-Xuan Ren;Chien-Chun Chen;Grant A. van Riessen;Philip Heng Wai Leong;Steve Feng Shu","doi":"10.1109/TCI.2025.3572250","DOIUrl":"https://doi.org/10.1109/TCI.2025.3572250","url":null,"abstract":"Ptychographic imaging confronts inherent challenges in applying deep learning for phase retrieval from diffraction patterns. Conventional neural architectures, both convolutional neural networks and Transformer-based methods, are optimized for natural images with Euclidean spatial neighborhood-based inductive biases that exhibit geometric mismatch with the concentric coherent patterns characteristic of diffraction data in reciprocal space. In this paper, we present PPN, a physics-inspired deep learning network with Polar Coordinate Attention (PoCA) for ptychographic imaging, that aligns neural inductive biases with diffraction physics through a dual-branch architecture separating local feature extraction from non-local coherence modeling. It consists of a PoCA mechanism that replaces Euclidean spatial priors with physically consistent radial-angular correlations. PPN outperforms existing end-to-end models, with spectral and spatial analysis confirming its greater preservation of high-frequency details. Notably, PPN maintains robust performance compared to iterative methods even at low overlap ratios — well-suited for high-throughput imaging in real-world acquisition scenarios for samples with consistent structural characteristics.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"11 ","pages":"888-900"},"PeriodicalIF":4.2,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11027575","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144557938","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
For bioluminescence tomography reconstruction, regularization algorithms and deep learning frameworks have been widely studied and achieved impressive results. However, the parameter selection of the regularization algorithm and the poor interpretability of deep learning methods have become the key factors that affect the reconstruction results and hinder its applicability. To mitigate the effects of this problem, in this paper, we proposed a novel residual graph model learning network (RGMLN) for bioluminescence tomography reconstruction by combining the advantages of regularization method and deep learning. RGMLN is based on the inference process of the thresholding iterative shrinkage algorithm. The difference is that the penalty term of the regularization method was replaced by a learnable nonlinear mapping between the residual and source distributions to ensure the interpretability of network. Meanwhile, considering the non-Euclidean property of the finite element mesh, a graph convolution operation based on Laplacian graph theory was conducted to aggregate features of mesh nodes using the topological information of the tetrahedral mesh. Lastly, based on residual learning and auto-encoder strategies, gradient descent and prox mapping modules were designed to structure the model-driven RGMLN method to take advantage of both the interpretability of iterative techniques and the flexibility of learning methods. Both numerical and in vivo experiments confirmed that the proposed network has excellent positioning accuracy and can be applied to different meshes and wavelengths.
{"title":"RGMLN:Residual Graph Model Learning Network for Bioluminescence Tomography","authors":"De Wei;Yizhe Zhao;Shuangchen Li;Heng Zhang;Beilei Wang;Xiaowei He;Jingjing Yu;Huangjian Yi;Xuelei He;Hongbo Guo","doi":"10.1109/TCI.2025.3572727","DOIUrl":"https://doi.org/10.1109/TCI.2025.3572727","url":null,"abstract":"For bioluminescence tomography reconstruction, regularization algorithms and deep learning frameworks have been widely studied and achieved impressive results. However, the parameter selection of the regularization algorithm and the poor interpretability of deep learning methods have become the key factors that affect the reconstruction results and hinder its applicability. To mitigate the effects of this problem, in this paper, we proposed a novel residual graph model learning network (RGMLN) for bioluminescence tomography reconstruction by combining the advantages of regularization method and deep learning. RGMLN is based on the inference process of the thresholding iterative shrinkage algorithm. The difference is that the penalty term of the regularization method was replaced by a learnable nonlinear mapping between the residual and source distributions to ensure the interpretability of network. Meanwhile, considering the non-Euclidean property of the finite element mesh, a graph convolution operation based on Laplacian graph theory was conducted to aggregate features of mesh nodes using the topological information of the tetrahedral mesh. Lastly, based on residual learning and auto-encoder strategies, gradient descent and prox mapping modules were designed to structure the model-driven RGMLN method to take advantage of both the interpretability of iterative techniques and the flexibility of learning methods. Both numerical and <italic>in vivo</i> experiments confirmed that the proposed network has excellent positioning accuracy and can be applied to different meshes and wavelengths.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"11 ","pages":"790-802"},"PeriodicalIF":4.2,"publicationDate":"2025-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144272944","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-28DOI: 10.1109/TCI.2025.3565138
Dan Pineau;François Orieux;Alain Abergel
The fusion of multispectral and hyperspectral data allows for restoring data with enhanced spatial and spectral resolutions. In cases of varying spatial blur, the current approach is to solve an ill-posed inverse problem by minimizing a mixed criterion. This minimization commonly involves an iterative gradient-based method. This paper proposes a new algorithm based on the Majorize-Minimize approach to compute the minimizer of a semi-quadratic convex edge-preserving criterion. The proposition relies on a reachable explicit solution of the quadratic majorant without the need to solve a Sylvester equation and for which we developed the proof of existence that was missing in a previous work. We conduct experiments on realistic synthetic measurements for the James Webb Space Telescope and show that our proposed solutions outperform the state-of-the-art in both computation time, achieving a 7000-fold speedup with the closed-form solution, and reconstruction quality, with a 2 dB PSNR improvement for the MM-based solution.
{"title":"Multispectral and Hyperspectral Image Fusion With Spectrally Varying Blurs and MM Algorithm","authors":"Dan Pineau;François Orieux;Alain Abergel","doi":"10.1109/TCI.2025.3565138","DOIUrl":"https://doi.org/10.1109/TCI.2025.3565138","url":null,"abstract":"The fusion of multispectral and hyperspectral data allows for restoring data with enhanced spatial and spectral resolutions. In cases of varying spatial blur, the current approach is to solve an ill-posed inverse problem by minimizing a mixed criterion. This minimization commonly involves an iterative gradient-based method. This paper proposes a new algorithm based on the Majorize-Minimize approach to compute the minimizer of a semi-quadratic convex edge-preserving criterion. The proposition relies on a reachable explicit solution of the quadratic majorant without the need to solve a Sylvester equation and for which we developed the proof of existence that was missing in a previous work. We conduct experiments on realistic synthetic measurements for the James Webb Space Telescope and show that our proposed solutions outperform the state-of-the-art in both computation time, achieving a 7000-fold speedup with the closed-form solution, and reconstruction quality, with a 2 dB PSNR improvement for the MM-based solution.","PeriodicalId":56022,"journal":{"name":"IEEE Transactions on Computational Imaging","volume":"11 ","pages":"704-716"},"PeriodicalIF":4.2,"publicationDate":"2025-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144171033","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：