Article
Computer Science, Software Engineering
Kunal Pradhan, Swarnajyoti Patra
Summary: Based on recent developments, the challenge in texture filtering lies in removing irregular and varying scale texture while preserving structural contents with minimal distortion. To address this issue, we propose a semantic-aware structure-preserving filtering technique. This technique generates an edge-map using semantic information extracted from global and local morphological gradient histograms, and then utilizes the edge-map to implement an adaptive median morpho-filtering with a dynamic window that avoids overlapping of textural and structural contents. Experimental results with various images demonstrate the potential of our proposed technique compared to state-of-the-art methods.
Article
Engineering, Electrical & Electronic
Lei He, Yongfang Xie, Shiwen Xie, Zhipeng Chen
Summary: This article proposes a structure-preserving texture smoothing method based on scale-aware bilateral total variation. The method utilizes a joint bilateral filter to construct the window bilateral variation and combines it with the window total variation to form a regularizer called bilateral total variation. This regularizer accurately quantifies the characteristics of texture and structure, allowing for the fine smoothing of salient textures while preserving weak edge structures and small structures. Experimental results demonstrate the superiority of the proposed method in texture smoothing and other applications.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Wei Liu, Pingping Zhang, Yinjie Lei, Xiaolin Huang, Jie Yang, Michael Ng
Summary: In this paper, a generalized smoothing framework is introduced using the truncated Huber penalty function, which provides strong flexibility. The framework can be applied to various image processing tasks and achieves superior performance in challenging cases.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Computer Science, Information Systems
Fernando J. Galetto, Guang Deng, Mukhalad Al-Nasrawi, Waseem Waheed
Summary: Edge-aware smoothing plays a crucial role in computer vision, graphics, and photography. This paper introduces a new and efficient local weighted average filter for edge-aware smoothing, which relies on linear filters only with computational complexity of O(N-pix). Statistical analysis and simulations provide insights into its computational efficiency and relationship with the bilateral filter. The proposed filter demonstrates comparable performance to state-of-the-art filters in various applications.
Article
Engineering, Electrical & Electronic
Lanling Zeng, Yucheng Chen, Yang Yang, Zhigeng Pan
Summary: In this paper, a novel weighted sparse gradient reconstruction model is proposed for edge-aware image smoothing. The proposed method suppresses low-amplitude gradients using an edge-aware mapping function and employs a weighted L1 gradient reconstruction model to obtain the output. The sparsity of the output gradients is enforced through L1 regularization, facilitating the edge-aware property. The weighted scheme enhances the edge awareness of the filter.
SIGNAL IMAGE AND VIDEO PROCESSING
(2023)
Article
Optics
Shuma Takeda, Mashiho Mukaida, Noriaki Suetake
Summary: In this letter, the authors propose a new edge-preserving smoothing filter called the cuboid filter. The cuboid filter calculates the mean of 3D neighboring pixels and achieves good performance in reducing small-amplitude noise and fast filtering. The effectiveness and validity of the cuboid filter are demonstrated through qualitative and quantitative evaluations.
Article
Computer Science, Artificial Intelligence
Yidan Feng, Sen Deng, Xuefeng Yan, Xin Yang, Mingqiang Wei, Ligang Liu
Summary: This article discusses whether a dataset's semantic description can assist a deep learning model in smoothing complex natural images. By generating ground-truth labels and synthesizing hard samples, a JESS-Net neural network is proposed along with using distinctive total variation loss to narrow the gap between synthetic and real data. Experiments demonstrate clear improvements in image cleanness and structure-preserving ability compared to state-of-the-art methods.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Engineering, Electrical & Electronic
Jun Li, Yuxuan Han, Yin Gao, Qiming Li, Sumei Wang
Summary: This paper proposes a new global optimization method that can preserve weak structures and smooth multiscale textures. The method separates textures and structures and gradually recovers shrinking edges/structures, demonstrating high effectiveness in maintaining weak structures and suppressing textures compared to current state-of-the-art methods.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Engineering, Electrical & Electronic
Hui Li, Tsz Nam Chan, Xianbiao Qi, Wuyuan Xie
Summary: This study proposes an improved multi-exposure fusion method that addresses the issue of detail loss by incorporating edge-preserving factors and a flexible bell curve function. Experimental results demonstrate that the method performs well in both static and dynamic scenes.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2021)
Article
Computer Science, Software Engineering
Dingkun Zhu, Weiming Wang, Xue Xue, Haoran Xie, Gary Cheng, Fu Lee Wang
Summary: Image smoothing is an important operation to enhance low-frequency structural parts and reduce noise and high-frequency textures in an image. This paper proposes a structure-preserving image smoothing network, which combines paired unsmoothed/smoothed images and meaningful edge information to improve performance. The network is trained using contrastive learning on the extended BSD500 dataset and utilizes an edge-aware total variation loss function for structure preservation. Experimental results show that our network outperforms state-of-the-art image smoothing methods in terms of SSIM and PSNR.
Article
Computer Science, Artificial Intelligence
Feihong Liu, Jun Feng, Geng Chen, Dinggang Shen, Pew-Thian Yap
Summary: Diffusion MRI magnitude data can be affected by noise floor, but can be avoided through phase correction. This study introduces an adaptive filtering approach called multi-kernel filter (MKF) which considers spatially-varying noise, showing significant improvements in spatial adaptivity and signal Gaussianization compared to state-of-the-art filters.
MEDICAL IMAGE ANALYSIS
(2021)
Article
Instruments & Instrumentation
Yunnan Xu, Pang Du, Ryan Senger, John Robertson, James L. Pirkle
Summary: In Raman spectroscopy, baseline correction is a critical step that has been recently improved with procedures relying on asymmetric loss functions. A novel baseline correction procedure called ISREA has been developed, utilizing smoothing splines to estimate the baseline, mimicking asymmetric square root loss, and avoiding direct optimization of a non-convex loss function by iteratively updating prediction errors and refitting baselines. Through extensive numerical experiments, ISREA has been shown to be simple, fast, and capable of yielding consistent and accurate baselines that preserve meaningful Raman peaks.
APPLIED SPECTROSCOPY
(2021)
Article
Engineering, Electrical & Electronic
Riya, Bhupendra Gupta, Subir Singh Lamba
Summary: This paper presents a novel structure-aware bilateral filter for texture smoothing while preserving structural information. The experimental results demonstrate the superiority of the proposed method in texture smoothing and structure preservation.
DIGITAL SIGNAL PROCESSING
(2022)
Article
Engineering, Electrical & Electronic
Lina Rong, Yuhan Kan, Xiangpeng Xie, Guo-Ping Jiang, Shengyuan Xu
Summary: This study focuses on the problem of edge-preserving consensus in discrete-time multi-agent systems. A method for designing channel filters, consisting of nominal elements and masking elements, is proposed using an edge-based parallel system design approach. Design criteria for edge-preserving protocols are provided based on the edge sensitivity matrix of the multi-agent system being studied. It is demonstrated that the design of masking elements is based on a single parameter of the nominal element and the H-infinity norm of a collapsed system. An algorithm for designing heterogeneous channel filters is also presented, using a cascade realization of several subsystems.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-EXPRESS BRIEFS
(2023)
Article
Engineering, Electrical & Electronic
Penghui Bu, Hong Zhao, Yusheng Jin, Yueyang Ma
Summary: In this paper, a novel O(N) time recursive non-local edge-aware filter is proposed, utilizing a unique graph and linear time algorithm to effectively propagate information across the entire image. Experimental results demonstrate the effectiveness of this filter in various computer vision and image processing applications.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2021)
Article
Environmental Sciences
Jinlei Ma, Zhiqiang Zhou, Bo Wang, Hua Zong, Fei Wu
Article
Engineering, Multidisciplinary
Changshen Xu, Lingjuan Miao, Zhiqiang Zhou, Yusen Lin
MEASUREMENT SCIENCE AND TECHNOLOGY
(2020)
Article
Engineering, Electrical & Electronic
Yusen Lin, Lingjuan Miao, Zhiqiang Zhou
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2020)
Article
Engineering, Electrical & Electronic
Yusen Lin, Lingjuan Miao, Zhiqiang Zhou, Changshen Xu
Summary: An error model for dual-axis RINS is established in this study, along with a new calibration method for nonorthogonal angles, which can accurately calibrate all angles. The observability analysis confirms that the proposed nonorthogonal angles are observable.
IEEE SENSORS JOURNAL
(2021)
Article
Environmental Sciences
Xiaowu Xiao, Bo Wang, Lingjuan Miao, Linhao Li, Zhiqiang Zhou, Jinlei Ma, Dandan Dong
Summary: This study addresses the issue of object detection in infrared and visible images by proposing a difference maximum loss function, as well as a focused feature-enhancement module and cascaded semantic extension module, which effectively improve detection performance.
Article
Environmental Sciences
Linhao Li, Zhiqiang Zhou, Bo Wang, Lingjuan Miao, Zhe An, Xiaowu Xiao
Summary: This paper proposes a domain adaptive ship detection method that minimizes domain discrepancies through image-level and instance-level adaption, enhancing the accuracy and generalization ability of ship detection.
Article
Environmental Sciences
Qi Ming, Lingjuan Miao, Zhiqiang Zhou, Junjie Song, Xue Yang
Summary: This paper proposes a sparse label assignment strategy to improve sample assignment in object detection, utilizes a position-sensitive feature pyramid network with a coordinate attention module to extract position-sensitive features, and introduces a distance rotated IoU loss to enhance bounding box regression.
Article
Geochemistry & Geophysics
Honghu Wang, Zhiqiang Zhou, Hua Zong, Lingjuan Miao
Summary: This paper proposes a novel wide-context attention network (W-CAN) that adaptively learns local features through two attention modules, enhancing feature performance with a hybrid loss and adding a branch to learn binary descriptors. Experiments show that this method outperforms some state-of-the-art RSIR methods on four benchmark data sets.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2021)
Article
Geochemistry & Geophysics
Qi Ming, Lingjuan Miao, Zhiqiang Zhou, Xue Yang, Yunpeng Dong
Summary: The proposed method optimizes bounding box regression for rotating objects in remote sensing images using representation invariance loss, Hungarian matching algorithm, and normalized rotation loss, achieving consistent and substantial improvement in localization accuracy.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Geochemistry & Geophysics
Qi Ming, Lingjuan Miao, Zhiqiang Zhou, Yunpeng Dong
Summary: This article discusses the role of discriminative features in object detection and proposes a critical feature capturing network (CFC-Net) to improve detection accuracy by building powerful feature representation, refining preset anchors, and optimizing label assignment.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Environmental Sciences
Xiaozhu Xie, Linhao Li, Zhe An, Gang Lu, Zhiqiang Zhou
Summary: This paper proposes a new method for detecting small ships in remote sensing images. The method generates region proposals using a hybrid anchor structure and incorporates strategies such as label reassignment, adaptive feature pyramid structure, and feature super-resolution to improve detection accuracy and reliability. Experimental results demonstrate the effectiveness of the proposed method.
Article
Engineering, Electrical & Electronic
Junjie Song, Lingjuan Miao, Qi Ming, Zhiqiang Zhou, Yunpeng Dong
Summary: In this article, we propose the RB-FPN and CSA label assignment strategy to address the challenges of fine-grained object detection in remote sensing images. RB-FPN fuses features from different layers and suppresses background information, providing high-quality semantic information for fine-grained object detection. The CSA label assignment strategy adaptively adjusts the IoU threshold based on statistical characteristics of oriented objects.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Yajun Qiao, Lingjuan Miao, Zhiqiang Zhou, Qi Ming
Summary: In this article, a novel detection method based on high-quality rotation proposal generation and adaptive angle optimization is proposed to address the issues caused by the variation in object shapes and orientations in remote sensing images. Experimental results show that the proposed method outperforms other methods in terms of average precision and has a simpler approach.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Junjie Song, Lingjuan Miao, Zhiqiang Zhou, Qi Ming, Yunpeng Dong
Summary: This letter addresses the problem of representing objects more appropriately in oriented object detection. The proposed point set distance (PSD) loss and probabilistic point set sample selection (PPSS) scheme effectively represent and select point sets of objects, achieving consistent and substantial improvements in arbitrary-oriented object detection, as demonstrated by experimental results.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Geochemistry & Geophysics
Linhao Li, Zhiqiang Zhou, Bo Wang, Lingjuan Miao, Hua Zong
Summary: This article proposes a novel CNN-based ship-detection method to overcome common deficiencies in current methods, aiming to more accurately locate ships in diverse orientations.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)