Article
Geochemistry & Geophysics
Weibo Huo, Xingyu Tuo, Yin Zhang, Yongchao Zhang, Yulin Huang
Summary: In this letter, an approach based on the balanced Tikhonov and total variation (TV) deconvolution is proposed for improving the azimuth resolution and obtaining the contour information of the target in radar forward-looking super-resolution imaging. The gradient function of the target scattering coefficient is used as the adaptive weighted parameter to automatically adjust the weighting between the penalty terms from TV and the Tikhonov regularization. The simulation and experimental results demonstrate the effectiveness of the proposed method, which outperforms traditional super-resolution imaging methods in outline retention.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Computer Science, Artificial Intelligence
Huan Wang, Yulun Zhang, Can Qin, Luc Van Gool, Yun Fu
Summary: This article presents a method called Global Aligned Structured Sparsity Learning (GASSL) to tackle the problem of efficient image super-resolution (SR). The method includes two major components: Hessian-Aided Regularization (HAIR) and Aligned Structured Sparsity Learning (ASSL). GASSL outperforms other recent methods in terms of efficiency, as demonstrated by extensive results.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Environmental Sciences
Qiping Zhang, Yin Zhang, Yongchao Zhang, Yulin Huang, Jianyu Yang
Summary: Total variation (TV) is an effective method to improve azimuth resolution and preserve contour information in airborne radar imaging, but with high computational complexity. A fast TV method based on Gohberg-Semencul (GS) representation is proposed in this paper to reduce complexity by utilizing low displacement rank feature of Toeplitz matrix. The proposed method is shown to improve resolution and preserve target contour efficiently in simulation and real data processing.
Article
Computer Science, Artificial Intelligence
Wende Dong, Shuyin Tao, Guili Xu, Yueting Chen
Summary: This paper proposes a regularized blind deconvolution method for restoring Poissonian blurred images, which achieves high quality restored images by combining L-0 norm, total variation, and negative logarithmic Poisson log-likelihood.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Geochemistry & Geophysics
Guo Xin, Gao Jian-Hu, Yin Xun-De, Yong Xue-Shan, Wang Hong-Qiu, Li Sheng-Jun
Summary: There is limited low-and-high frequency information in seismic data, resulting in lower seismic resolution. This study proposes a novel method to improve seismic resolution by using expected wavelet spectrum in the frequency domain and Frobenius vector regularization of the Hessian matrix in the time domain. It effectively improves the prediction accuracy of thin reservoirs and thin interbeds.
APPLIED GEOPHYSICS
(2022)
Article
Engineering, Electrical & Electronic
Wende Dong, Xiaoyan Xu, Chenlong Zhu, Luqi Hu, Guili Xu, Shuyin Tao
Summary: We propose a dehazing problem model with hybrid regularization and design an effective algorithm to restore the latent image and transmission map simultaneously. Experimental results show that our approach can achieve a high-quality restored image.
JOURNAL OF ELECTRONIC IMAGING
(2022)
Article
Environmental Sciences
Jie Han, Songlin Zhang, Shouzhu Zheng, Minghua Wang, Haiyong Ding, Qingyun Yan
Summary: This paper focuses on the application of sparsity regularization based on the L-1 norm in solving ill-posed sparsity inversion problems. It proposes a partially bias-corrected solution to improve the rigor of the theory. Experimental results demonstrate that the proposed method with partial bias correction achieves higher quality compared to without bias correction.
Article
Engineering, Electrical & Electronic
Shubhabrata Sarkar, Pankaj Wahi, Prabhat Munshi
Summary: The study focuses on improving the performance of a table-top CT scanning instrument by proposing a new super resolution technique based on higher order total variation minimization. Experimental results show the technique can produce high resolution images from low resolution ones and outperforms existing super resolution techniques. The proposed technique is validated with experiments on different objects and shows potential as a cost-effective solution to resolution problems caused by detector size constraints in table-top CT scanning instruments.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Engineering, Multidisciplinary
Zuoxun Tan, Hu Yang
Summary: Motivated by the superior performance of nonconvex nonsmooth L-p (0 < p < 1) norm, this paper introduces a novel method that combines the weighted Schatten p-norm, L-p-norm, and total variation regularization based on the multiple matrices denoising framework. An efficient alternating direction method of multipliers (ADMM) is designed to solve the nonconvex and nonsmooth model. Extensive experiments on face datasets, videos, and real-world noisy images demonstrate that the proposed method significantly improves denoising performance, particularly for removing large sparse noise.
APPLIED MATHEMATICAL MODELLING
(2023)
Article
Computer Science, Artificial Intelligence
Razieh Sheikhpour, Kamal Berahmand, Saman Forouzandeh
Summary: Feature selection aims to eliminate redundant features and choose informative ones. Semi-supervised feature selection becomes important as it utilizes labeled and unlabeled data. We propose two frameworks, one based on Hessian matrix and the other on Hessian-Laplacian combination, for semi-supervised feature selection. Our frameworks utilize regularization and constraint techniques to select informative features and maintain the topological structure of data. Experimental results demonstrate the effectiveness of our frameworks in selecting informative features.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Mathematics, Applied
Baoli Shi, Fang Gu, Zhi-Feng Pang, Yuhua Zeng
Summary: This paper proposes a new model for removing salt and pepper noise by combining high order total variation regularization with nuclear norm regularization. The model is convex and separable, and the classic alternating direction method of multipliers is used to solve it. The experimental comparisons demonstrate that the proposed model outperforms other methods in terms of signal-to-noise ratio (SNR) and structural similarity index (SSIM).
APPLIED MATHEMATICS AND COMPUTATION
(2022)
Article
Automation & Control Systems
Mourad Nachaoui, Amine Laghrib
Summary: This article proposes an improved super-resolution method that combines bilevel optimization technique with parameter learning strategy to effectively remove Gaussian and speckle noises, thus improving the quality of the restored image.
JOURNAL OF THE FRANKLIN INSTITUTE-ENGINEERING AND APPLIED MATHEMATICS
(2022)
Article
Statistics & Probability
Mariano Gabitto, Herve Marie-Nelly, Ari Pakman, Andras Pataki, Xavier Darzacq, Michael Jordan
Summary: This study addresses the challenge of single-molecule identification in super-resolution microscopy using a Bayesian nonparametric framework. The algorithm shows promising performance in localizing molecules with known spatial positions.
ANNALS OF APPLIED STATISTICS
(2021)
Article
Chemistry, Analytical
Jongeun Park, Hansol Kim, Moon Gi Kang
Summary: To improve the performance of super-resolution algorithms, various SR kernel models have been proposed. However, they lead to unpleasant artifacts in the output images. KernelGAN, a conventional research, introduces GANs to estimate SR kernels from single images. However, it still faces challenges in estimating large-sized and anisotropic kernels due to the insufficient consideration of structural information. Therefore, this paper proposes a kernel estimation algorithm called TVG-KernelGAN, which efficiently focuses on the structural information of input images. Experimental results show that the proposed method accurately and stably estimates kernels and improves the performance of super-resolution algorithms.
Article
Environmental Sciences
Xiaolong Chen, Wei Xu, Shuping Tao, Tan Gao, Qinping Feng, Yongjie Piao
Summary: This paper introduces the critical technology of infrared dim small target detection and proposes a novel method based on total variation weighted low-rank constraint to solve the staircase effect problem in target detection. The proposed method shows improved detection accuracy and stronger robustness under complex background conditions.
