Article
Computer Science, Artificial Intelligence
Yongxiang Yao, Yongjun Zhang, Yi Wan, Xinyi Liu, Xiaohu Yan, Jiayuan Li
Summary: Traditional image feature matching methods are not satisfactory for multi-modal remote sensing images due to nonlinear radiation distortion differences and complicated geometric distortion. This paper proposes a new robust MRSI matching method based on co-occurrence filter space matching, which optimizes the matching by constructing a new co-occurrence scale space, extracting feature points, and optimizing the distance function. Experimental results show that the proposed method significantly outperforms other state-of-the-art methods in terms of matching effectiveness.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Engineering, Electrical & Electronic
Jiaxuan Chen, Shuang Chen, Yuyan Liu, Xiaoxian Chen, Yang Yang, Yungang Zhang
Summary: This article proposes a method based on local structure visualization descriptors and convolutional neural networks for image registration, aiming to improve the reliability and precision of feature matching. The method is not restricted by specific transformation models, has good adaptability to various remote sensing images, and extensive experiments demonstrate its superior performance.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Engineering, Biomedical
Sajib Saha, G. M. Atiqur Rahaman, Tazul Islam, Masuma Akter, Shaun Frost, Yogesan Kanagasingam
Summary: Registration of retinal image is a crucial step in medical diagnoses, and this paper proposes an innovative method that utilizes log-polar transform and a novel descriptor CLHB for precise matching. Experimental results show that the proposed method outperforms existing state-of-the-art methods on both public and private datasets, achieving higher accuracy by 2-3%.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2021)
Article
Engineering, Electrical & Electronic
Li Xue, Yehua Sheng, Ka Zhang
Summary: This study proposes a robust descriptor construction algorithm to improve the matching problem in heterogeneous optical remote sensing images by considering the internal gray value changes of ground objects. Experimental results demonstrate that the proposed algorithm exhibits better stability and capability in matching homologous and heterogeneous images compared to other commonly used algorithms.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Computer Science, Artificial Intelligence
Zhanlong Yang, Linzhi Yang, Geng Chen, Pew-Thian Yap
Summary: In this paper, a novel local image descriptor called IPCET is proposed, which is based on the phase and amplitude information of PCET. The IPCET descriptor is robust to both photometric and geometric transformations, and outperforms cutting-edge moment-based descriptors according to extensive experiments.
PATTERN RECOGNITION
(2023)
Article
Geochemistry & Geophysics
Qiang Xiong, Shenghui Fang, Xiaojuan Liu, Xueqin Jiang, Yongfei Lv
Summary: This letter proposes a novel descriptor for multispectral image matching, which combines phase consistency gradient and log-polar coordinates to eliminate nonlinear radiation differences and extract more robust features. Experimental results demonstrate that this method outperforms traditional feature matching methods.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Computer Science, Artificial Intelligence
Shanshan Zhao, Mingming Gong, Haimei Zhao, Jing Zhang, Dacheng Tao
Summary: Recent studies have achieved promising results by jointly learning local feature detectors and descriptors. To overcome the lack of ground-truth keypoint supervision, previous methods have incorporated relevant knowledge about keypoint attributes into the network for enhanced model learning. This paper presents Deep Corner, an end-to-end deep network that combines a local similarity-based keypoint measure with a plain convolutional network, inspired by traditional corner detectors. The proposed method yields reliable keypoints, facilitate the learning of distinctive descriptors. Additionally, the paper introduces a multi-level U-Net architecture and a feature self-transformation operation to further improve keypoint localization and descriptor invariance.
INTERNATIONAL JOURNAL OF COMPUTER VISION
(2023)
Article
Engineering, Electrical & Electronic
Carlos Eduardo Padilla Leyferman, Jose Trinidad Guillen Bonilla, Juan Carlos Estrada Gutierrez, Maricela Jimenez Rodriguez
Summary: Facial recognition is currently of great importance in authentication processes to prevent unauthorized access. This paper proposes a new texture descriptor called Cyclical Chroma and a new classification technique considering sub-pixel values of each RGB channel. Tests were conducted on controlled and uncontrolled image databases, demonstrating the effectiveness of the proposed techniques with 100% efficiency under controlled conditions and 78% effectiveness under uncontrolled conditions before equalization, which improved efficiency to 100%.
IET COMMUNICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Ignacio Rocco, Mircea Cimpoi, Relja Arandjelovic, Akihiko Torii, Tomas Pajdla, Josef Sivic
Summary: In this work, we propose a convolutional neural network architecture that identifies sets of spatially consistent matches by analyzing the neighbourhood consensus patterns in the 4D space of all possible correspondences between a pair of images without the need for a global geometric model. The model can be effectively trained from weak supervision and achieves state-of-the-art results on various matching tasks.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Chemistry, Multidisciplinary
Liang Tang, Shuhua Ma, Xianchun Ma, Hairong You
Summary: This paper proposes an improved SIFT algorithm with an added stability factor for image feature matching, which reduces matching time and algorithm error.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Information Systems
Yuan Wang, Xiangyang Liang, Lei Chen
Summary: This study proposes a novel image registration algorithm that addresses the problem of low registration accuracy caused by resolution and spectral differences between infrared and visible images. The algorithm includes steps such as image translation, incorporation of a normalization-based attention mechanism, introduction of a hybrid loss function, guided filtering, and the use of Speeded-Up Robust Features (SURF) and random sample consensus (RANSAC) algorithms. Experimental results showed an improved registration accuracy of approximately 5% compared to LNIFT algorithms.
Article
Engineering, Electrical & Electronic
Tiecheng Song, Jie Feng, Lin Luo, Chenqiang Gao, Hongliang Li
Summary: In this paper, two novel operators, local grouped order pattern (LGOP) and non-local binary pattern (NLBP), are proposed for texture description. Experimental results demonstrate that combining LGOP and NLBP to construct discriminative histogram features as texture descriptor LGONBP shows superiority over state-of-the-art LBP variants for texture classification.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2021)
Article
Environmental Sciences
Kun Yu, Chengcheng Xu, Jie Ma, Bin Fang, Junfeng Ding, Xinghua Xu, Xianqiang Bao, Shaohua Qiu
Summary: This paper proposes a novel automatic matching method named LURF for multimodal images by extracting semantic road features, designing an intersection point detector, and a local entropy descriptor, and adopting a global optimization strategy for correct matching.
Article
Engineering, Multidisciplinary
Quan Wu, Zhenhua Li, Shipeng Zhu, Peng Peng Xu, Ting Ting Yan, Junpu Wang
Summary: This paper proposes a novel feature descriptor HMPC based on the structural properties of images, with a new distance formula designed for calculating similarity. The method shows robust and accurate matching performance compared to state-of-the-art methods when applied to multi-source image matching tasks.
Article
Engineering, Electrical & Electronic
Shuo Li, Xiaolei Lv, Hao Wang, Jian Li
Summary: In this article, a novel fast, robust, and extensible matching method based on the primary structure-weighted orientation consistency (PSOC) is proposed to address the challenging task of matching multimodal remote sensing images. The method extracts consistent primary structures and suppresses texture details effectively, leading to improved matching success rate and reduced complexity. Experimental results demonstrate the superiority of the proposed method over state-of-the-art descriptors.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Siyuan Zou, Xinyi Liu, Xu Huang, Yongjun Zhang, Senyuan Wang, Shuang Wu, Zhi Zheng, Bingxin Liu
Summary: This letter presents a LiDAR and image line-guided stereo matching method (L2GSM) that combines sparse but high-accuracy LiDAR points and sharp object edges of images to produce accurate and detailed point clouds. By using LiDAR depth information to extract depth discontinuity lines on the image, trilateral update of cost volume and depth discontinuity lines-aware semi-global matching (SGM) strategies are proposed to incorporate LiDAR data and depth discontinuity lines into the dense matching algorithm. Experimental results on indoor and aerial datasets demonstrate that our method greatly enhances the performance of the original SGM and surpasses two state-of-the-art LiDAR constraints' SGM methods, particularly in recovering the 3-D structure of low-textured and depth discontinuity regions. Moreover, the 3-D point clouds generated by our proposed method outperform the LiDAR data and dense matching point clouds generated by Metashape and SURE aerial in terms of completeness and edge accuracy.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Geochemistry & Geophysics
Bin Zhang, Yongjun Zhang, Yansheng Li, Yi Wan, Yongxiang Yao
Summary: In this study, a lightweight vision transformer network called CloudViT is proposed for cloud detection from satellite imagery. By using dark channel priors in multispectral imagery to guide the network to learn features, the network is able to enhance image features and produce more accurate cloud detection results. Additionally, a plug-and-play channel adaptive module is introduced to address the inconsistency in the number of bands from different satellite sensors.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Computer Science, Artificial Intelligence
Yansheng Li, Fanyi Wei, Yongjun Zhang, Wei Chen, Jiayi Ma
Summary: This paper proposes a novel hierarchical spectral and structure-preserving fusion network (HS2P) to recover cloud and shadow regions in optical remote sensing imagery based on the hierarchical fusion of optical and SAR remote sensing imagery. Extensive experiments demonstrate that the proposed method achieves significant improvements in recovering diverse ground information in optical remote sensing imagery with various cloud types.
INFORMATION FUSION
(2023)
Article
Geography, Physical
Yongjun Zhang, Yongxiang Yao, Yi Wan, Weiyu Liu, Wupeng Yang, Zhi Zheng, Rang Xiao
Summary: A new multi-modal remote sensing image matching method called Histogram of the Orientation of Weighted Phase (HOWP) is proposed in this paper. This method improves the performance of remote sensing image matching by optimizing feature points, establishing a weighted phase orientation model, and constructing a regularization-based log-polar descriptor. Experimental results show that the proposed method performs better in resisting radiometric distortion and contrast differences, and effectively tackles the problems of direction reversal and phase extreme value mutation.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2023)
Article
Geography, Physical
Xinyi Liu, Xianzhang Zhu, Yongjun Zhang, Senyuan Wang, Chen Jia
Summary: In this paper, a novel method for generating concise building models from dense meshes is proposed. The method involves extracting and completing planar primitives of the building, and reconstructing a concise polygonal mesh through connectivity-based primitive assembling. The method demonstrates high efficiency and robustness.
PHOTOGRAMMETRIC RECORD
(2023)
Article
Geography, Physical
Shunping Ji, Chang Zeng, Yongjun Zhang, Yulin Duan
Summary: This paper comprehensively evaluates the performance of conventional and deep learning-based image-matching methods in the field of remote sensing. The results show that the combination of SIFT, ContextDesc, and NNDR achieves the best results when using more comprehensive indicators, and is recommended for use in remote sensing.
PHOTOGRAMMETRIC RECORD
(2023)
Article
Geography, Physical
Daifeng Peng, Chenchen Zhai, Yongjun Zhang, Haiyan Guan
Summary: This paper proposes a network based on dense connections and attention feature fusion for ground object change detection. It effectively solves the problems of incomplete information and inaccurate edges, and achieves the best performance in both quantitative and qualitative evaluations.
PHOTOGRAMMETRIC RECORD
(2023)
Article
Engineering, Electrical & Electronic
Bin Zhang, Yongjun Zhang, Yansheng Li, Yi Wan, Haoyu Guo, Zhi Zheng, Kun Yang
Summary: In this study, a new semisupervised deep semantic labeling framework is proposed for the semantic segmentation of high-resolution RS images. The model uses transformation consistency regularization to encourage consistent network predictions under different random transformations or perturbations. Three different transforms were tried to compute the consistency loss and analyze their performance. A comprehensive experiments on two RS datasets confirmed that the suggested approach utilized latent information from unlabeled samples to obtain more precise predictions and outperformed existing semisupervised algorithms in terms of performance. Our experiments further demonstrated that our semisupervised semantic labeling strategy has the potential to partially tackle the problem of limited labeled samples for high-resolution RS image land-cover segmentation.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Engineering, Electrical & Electronic
Yongjun Zhang, Yameng Wang, Yi Wan, Wenming Zhou, Bin Zhang
Summary: In this study, we propose PointBoost, an effective sequential segmentation framework that can directly process cross-modal data of LiDAR point clouds and imagery, extracting richer semantic features from cross-dimensional and cross-modal information. Ablation experiments demonstrate that PointBoost takes full advantage of the 3-D topological structure between points and attribute information of point clouds, which is often discarded by other methods. Experiments on three multimodal datasets show that our method achieves superior performance with good generalization.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Zhi Gao, Wenbo Sun, Yao Lu, Yichen Zhang, Weiwei Song, Yongjun Zhang, Ruifang Zhai
Summary: In this article, a deep multitask learning framework is proposed to improve the performance of semantic segmentation (SS) and height estimation (HE) tasks in remote sensing scene understanding. Two novel objective functions, cross-task contrastive (CTC) loss and cross-pixel contrastive (CPC) loss, are introduced to enhance the performance through contrastive learning. Experimental results demonstrate that the proposed approach significantly outperforms the state-of-the-art methods in both SS and HE.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Wanshan Peng, Yan Gong, Shenghui Fang, Yongjun Zhang, Jadunandan Dash, Jie Ren, Jiacai Mo
Summary: This article proposes a radiometric block adjustment model considering the vignetting effect and the light-dark differences between images. The method requires only a small number of calibration samples, reducing the complexity of the experiment. The results show that this method can compensate for vignetting to some extent and improve the radiometric consistency of the dataset.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Remote Sensing
Senyuan Wang, Xinyi Liu, Yongjun Zhang, Jonathan Li, Siyuan Zou, Jipeng Wu, Chuang Tao, Quan Liu, Guorong Cai
Summary: We propose a semantic-guided building reconstruction method called SGR, which can achieve independent and complete reconstruction of building models. The method consists of two key stages: 2.5D convex cell complex representation for space partition and semantic-guided graph-cut formulation to eliminate interference. Experimental results show that SGR can authentically reconstruct weakly observed surfaces and obtain watertight models considering fidelity, integrity, and time complexity.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)
Article
Remote Sensing
Xiaojian Liu, Yongjun Zhang, Huimin Zou, Fei Wang, Xin Cheng, Wenpin Wu, Xinyi Liu, Yansheng Li
Summary: In this study, we proposed a novel method to detect marine oil spills by constructing a multi-source knowledge graph. Our method effectively organizes and utilizes various oil spill-related information and selects favorable features for oil spill detection. By combining rule inference and graph neural network technology, our method shows high sensitivity, specificity, and precision in identifying oil spills even in severely imbalanced data conditions.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)
Article
Remote Sensing
Yongjun Zhang, Yixin Lu, Yansong Duan, Dong Wei, Xianzhang Zhu, Bin Zhang, Bohui Pang
Summary: Crack detection plays a key role in civil engineering, and vision-based methods are widely used. The balance between global and local information is crucial in detecting cracks from different sources. Many existing methods focus on crack detection in handheld photographs and may not perform well on UAV-generated images or images with different backgrounds. To address this challenge, we propose a robust and innovative method called Crack Detection with Structure Line (CDSL).
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)