Article
Environmental Sciences
Ying Zhu, Tingting Yang, Mi Wang, Hanyu Hong, Yaozong Zhang, Lei Wang, Qilong Rao
Summary: This paper proposes a jitter detection method using sequence CMOS images for high-resolution remote sensing satellites, which accurately measures jitter errors and recovers relevant information.
Article
Optics
Haiqiu Liu, Huimin Ma, Qixing Tang, Dong Wang
Summary: This research investigates the noise-amplifying problems in jitter detection, focusing on determining the error transfer coefficients between jitter and offset, and categorizing frequencies into blind, noise-amplifying, and noise-suppressing categories. Formulas are established to determine blind frequencies and noise-amplifying bands for two adjacent CCDs, as well as to prove the inevitability of aliasing between noise-amplifying bands of three adjacent CCDs forming two CCD pairs. The experiments and simulations conducted show that the established formulas can generate reliable results for detecting jitter components through analyzing CCD parameters.
OPTICS COMMUNICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Zhi-Ze Wu, Xiao-Feng Wang, Le Zou, Li-Xiang Xu, Xin-Lu Li, Thomas Weise
Summary: Object detection from satellite images is challenging due to large coverage areas and objects of different scales. A hierarchical framework with deep feature extraction and inclined bounding box mechanism is proposed to improve efficiency. Experimental results demonstrate the superior performance of the proposed method compared to standalone state-of-the-art object detectors.
APPLIED SOFT COMPUTING
(2021)
Article
Geochemistry & Geophysics
Jun Pan, Guo Ye, Ying Zhu, Xiaolin Song, Fen Hu, Chi Zhang, Mi Wang
Summary: This article proposes a continuous dynamic shooting model (CDSM) and a method based on jitter detection and image restoration for TDI CCDs, aiming to improve the accuracy of jitter detection and enhance the quality of restored images. The method utilizes an integral transformation function and adaptive image restoration based on context to achieve improved results in detecting jitter and enhancing the geometric and radiometric qualities of restored images.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Multidisciplinary Sciences
Chunmei Wang, Xingfa Gu, Xiang Zhou, Jian Yang, Tao Yu, Zui Tao, Hailiang Gao, Yulin Zhan, Xiangqin Wei, Juan Li, Lili Zhang, Lei Li, Bingze Li, Zhuangzhuang Feng, Xigang Wang, Ruoxi Fu, Xingming Zheng, Chunnuan Wang, Yuan Sun, Bin Li, Wen Dong
Summary: The SONTE-China network in China measures pixel- and multilayer-based soil moisture and temperature, covering dry and wet zones. It validates soil moisture products and provides basic data for weather forecasting, flood forecasting, agricultural drought monitoring, and water resource management.
Article
Geochemistry & Geophysics
Fangrong Zhou, Weishi Jin, Zezhong Zheng, Fan Mou, Zhongnian Li, Yutang Ma, Bu Wei, Shuangde Huang, Qun Wang
Summary: This article proposes a novel method for detecting insulators on 500-kV transmission towers in remote sensing images. The method consists of three components: a super-resolution network, an object detection model, and a semantic segmentation network. Experimental results show that the proposed method effectively detects insulators in high-resolution satellite images and achieves the highest F1 score of 0.7952.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Environmental Sciences
Aida Alvera-Azcarate, Dimitry Van der Zande, Alexander Barth, Joao Felipe Cardoso dos Santos, Charles Troupin, Jean-Marie Beckers
Summary: This method proposes a way to detect cloud shadows over the ocean by applying a series of tests. It is not dependent on the wavebands measured by a specific satellite sensor, and works with cloud shadows of all sizes, including very small object shadows.
REMOTE SENSING OF ENVIRONMENT
(2021)
Article
Engineering, Electrical & Electronic
Chujie Wei, Tao Fang, Yingle Fan, Wei Wu, Ming Meng, Qingshan She
Summary: This study proposes a new method of image contour detection based on binocular parallax, which can extract the primary contour of an image and suppress local texture through innovative adjustments of opponent cell connection weights, binocular parallax energy model, and multi-scale receptive field fusion strategy. It provides a new idea for subsequent studies on the higher visual cortex's image understanding and visual cognition.
SIGNAL IMAGE AND VIDEO PROCESSING
(2022)
Article
Environmental Sciences
Chandi Witharana, Mahendra R. Udawalpola, Anna K. Liljedahl, Melissa K. Ward Jones, Benjamin M. Jones, Amit Hasan, Durga Joshi, Elias Manos
Summary: This study aims to develop a deep learning convolutional neural network model for automatically detecting and characterizing retrogressive thaw slumps in the Arctic, and investigate the impact of input image tile size and resizing factor on detection accuracy, as well as the transferability of the model across different training samples.
Article
Engineering, Electrical & Electronic
Kenichi Sasaki, Tatsuyuki Sekine, Louis-Jerome Burtz, William J. Emery
Summary: This study uses satellite images and machine learning techniques to estimate the amount and type of marine debris on the beaches of southern Japan. The results show that Shannon's entropy computed from World-View 2 and 3 imagery can effectively detect and map coastal debris, even in areas without ground truth data.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Environmental Sciences
Junghoon Seo, Wonkyu Park, Taejung Kim
Summary: This paper proposes a new approach to detecting small-object changes in high-resolution satellite images. The approach uses feature points extracted from previous and recent images to estimate the magnitude of change and reduce false alarms. Experimental results show that this feature-based approach outperforms pixel-based methods in terms of precision and accuracy.
Article
Engineering, Electrical & Electronic
Bo Guo, Ruixiang Zhang, Haowen Guo, Wen Yang, Huai Yu, Peng Zhang, Tongyuan Zou
Summary: Fine-grained ship detection is a challenging task due to large aspect ratios and severe category imbalance. To address this, we propose a shape-aware feature learning method to mitigate misalignments and a shape-aware instance switching method to balance category distribution. Experimental results on our multicategory ship detection dataset demonstrate the superiority of our proposed method over state-of-the-art methods.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Geography, Physical
Xuan Hou, Yunpeng Bai, Ying Li, Changjing Shang, Qiang Shen
Summary: This paper proposes a High-Resolution Triplet Network (HRTNet) framework to address the issue of temporal information missing in deep learning-based change detection in remote sensing images. By introducing a novel triplet input network and dynamic inception module, the effectiveness and robustness of change detection are successfully improved.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2021)
Article
Environmental Sciences
Zhaofei Liu
Summary: This study comprehensively evaluated satellite precipitation products (SPPs) in the Qinghai-Tibetan Plateau region and found that GPM and MSWEP performed the best in terms of precipitation frequency and series. SPPs showed good performance in detecting precipitation occurrence but relatively weak performance in measuring precipitation series. The evaluation results showed better performance in monsoon-affected regions and poorer performance in westerly circulation areas. Multiple indices representing different characteristics are recommended for a comprehensive evaluation of SPPs.
Article
Geochemistry & Geophysics
Shijie Liu, Han Zhang, Xiaohua Tong, Zhen Ye, Huan Xie, Feng Lin
Summary: This letter proposes a scheme to investigate the effect of matching window size and TDI stage number on image-based jitter detection for time-delay integration (TDI) linear push-broom satellite images. The model of image motion caused by jitter and the model of jitter detection and estimation for TDI images are derived and established. Experimental analysis reveals that the influence of TDI stage number and matching window size on jitter frequency detection is not significant. However, the matching window size in the row direction has a significant effect on jitter amplitude detection.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Engineering, Electrical & Electronic
Zhonghua Hong, Changyou Xu, Xiaohua Tong, Shijie Liu, Ruyan Zhou, Haiyan Pan, Yun Zhang, Yanling Han, Jing Wang, Shuhu Yang
Summary: This paper proposes a method that can optimize the luminance, contrast, and color difference of remote-sensing images. By processing the chrominance and luminance channels of the image in the YCbCr color space, it reduces the influence of different channels. The proposed method has been tested on challenging datasets and shows better results than state-of-the-art approaches in terms of visuals and quality indicators.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Engineering, Electrical & Electronic
Yun Zhang, Qi Lu, Qin Jin, Wanting Meng, Shuhu Yang, Shen Huang, Yanling Han, Zhonghua Hong, Zhansheng Chen, Weiliang Liu
Summary: This article introduces two different SSH inversion models based on CYGNSS, using BP neural network and CNN methods, and evaluates their performance on different training sets and verification data.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Environmental Sciences
Hua Yang, Ming Chen, Guowen Wu, Jiali Wang, Yingxi Wang, Zhonghua Hong
Summary: Hyperspectral data consists of numerous narrow spectral bands, providing more detailed spectral characteristics than multispectral data. However, the correlated bands in hyperspectral data result in computational complexity, limiting their applications. Dimensionality reduction of hyperspectral data is crucial for analysis, and deep reinforcement learning (DRL) is used for band selection. This study proposes a new reward mechanism strategy and utilizes Double Deep Q-Network (DDQN) in DRL to improve stability and accuracy.
Article
Green & Sustainable Science & Technology
Kuifeng Luan, Zhaoxiang Cao, Song Hu, Zhenge Qiu, Zhenhua Wang, Wei Shen, Zhonghua Hong
Summary: Statistical analysis of CALIPSO L3 data from 2007 to 2020 was conducted to investigate the horizontal and vertical distributions of aerosol properties in the Taklimakan Desert, North central region of China, North China Plain, and Yangtze River Delta. The study aimed to identify similarities and differences in atmospheric aerosols in different regions and evaluate the impact of pollution control policies in China on aerosol properties. Results show that aerosol optical depth (AOD) had high annual averages in the study areas, but exhibited a decline after the implementation of pollution control policies. The aerosol extinction coefficient showed clear regional patterns and a tendency to decrease with increasing altitude.
Article
Geography, Physical
Shenlu Jiang, Yuliya Tarabalka, Wei Yao, Zhonghua Hong, Guofu Feng
Summary: This study focuses on accelerating the processing of deep neural networks (DNNs) on high-resolution aerial images by optimizing the system architecture. Parallel processing and a three-level memory system were proposed to reduce data transfer and improve training efficiency. Experimental results showed that the proposed system significantly improved prediction speed without sacrificing accuracy.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2023)
Article
Geography, Physical
Qing Fu, Xiaohua Tong, Shijie Liu, Zhen Ye, Yanmin Jin, Hanyu Wang, Zhonghua Hong
Summary: In this paper, a combined Preconditioned Conjugate Gradient (PCG) and Graphic Processing Unit (GPU) parallel computing approach is proposed to improve the efficiency of large-scale bundle adjustment (BA) of high-resolution satellite imagery (HRSI) without ground control points (GCPs). The experimental results show that the proposed method has lower memory consumption, higher computational efficiency, and comparable accuracy compared to other methods.
PHOTOGRAMMETRIC ENGINEERING AND REMOTE SENSING
(2023)
Article
Instruments & Instrumentation
Ruyan Zhou, Jinmeng Gao, Zhonghua Hong, Haiyan Pan, Yun Zhang, Yanling Han, Jing Wang, Shuhu Yang
Summary: Gaofen-7 (GF-7) is equipped with a laser altimetry system to improve its elevation positioning accuracy. The absence of this system may limit GF-7's accuracy. To address this, a method of combining external laser altimetry points (LAPs) with GF-7 stereo images was proposed. The method effectively improved registration accuracy and decreased the root mean square error (RMSE) of elevation positioning from about 5.5 to 1.3 m.
SENSORS AND MATERIALS
(2023)
Article
Engineering, Electrical & Electronic
Shenlu Jiang, Zhonghua Hong
Summary: In this study, a system for detecting unexpected dynamic obstacles is built by combining an understanding of a road scene, optical flow movement tracking, and low-cost online visual tracking. The system is able to rapidly track pixel flows and detect targets, while efficiently allocating GPU and CPU resources in real-time. Experimental results demonstrate that the system achieves high efficiency and accuracy in urban road scenes and is capable of handling complex indoor environments.
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE
(2023)
Article
Geography, Physical
Zhonghua Hong, Changyou Xu, Xiaohua Tong, Shijie Liu, Ruyan Zhou, Haiyan Pan, Yun Zhang, Yanling Han, Jing Wang, Shuhu Yang
Summary: An improved method for optimizing color consistency across multiple images is proposed, which utilizes optimized low-resolution reference images to enhance image quality. The method includes reconstructing affected areas, minimizing color differences, smoothing the image boundary, ensuring color continuity, and correcting image color based on optimized reference and down-sampled images. The approach shows significant advantages in both quantitative indicators and visual quality compared to state-of-the-art methods.
GISCIENCE & REMOTE SENSING
(2023)
Article
Computer Science, Artificial Intelligence
Shenlu Jiang, Runze Cui, Runze Wei, Zhiyang Fu, Zhonghua Hong, Guofu Feng
Summary: Person-following is crucial for service robots, and vision technology is the leading trend for building environmental understanding. Our novel approach combines real-time tracking-by-segmentation and future motion estimation to overcome the limitations of existing methodologies. We validate our method using multiple tracking datasets and demonstrate its effectiveness for long-term person-following tasks in indoor environments, with promise for practical implementation in service robots.
FRONTIERS IN NEUROROBOTICS
(2023)
Article
Engineering, Electrical & Electronic
Zhonghua Hong, Linxuan Zhong, Xiaohua Tong, Haiyan Pan, Ruyan Zhou, Yun Zhang, Yanling Han, Jing Wang, Shuhu Yang, Haiyang He
Summary: This article presents an enhanced double-branch network algorithm for estimating crater ages using semisupervised learning and multisource lunar data. The algorithm consists of three steps: semisupervised training data augmentation, adaptive two-branch feature extraction, and a two-stage crater age classification process. The improved approach achieved an overall accuracy of 83.7% on the test set of meteorite craters, which is 5.2% higher than the previous deep learning method.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Engineering, Electrical & Electronic
Ruyan Zhou, Shangcheng Hu, Zhonghua Hong, Xiaohua Tong, Shijie Liu, Haiyan Pan, Yun Zhang, Yanling Han, Jing Wang, Shuhu Yang
Summary: This study proposes an adaptive calculation method based on different GPU performance parameters and automatic image blocking to accelerate the orthorectification process. Experimental results show that the adaptive calculation method achieves a 43.22% higher acceleration rate and a 34.41 times faster correction time compared to the general GPU for a single ZY-3 image. For large-batch images, the adaptive GPU achieves a 32.6% higher acceleration rate than the general GPU, providing an adaptive optimization strategy for large-batch image orthorectification.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Computer Science, Information Systems
Zhonghua Hong, Shihao Sun, Xiaohua Tong, Ruyan Zhou, Haiyan Pan, Yun Zhang, Yanling Han, Jing Wang, Shuhu Yang, Zhenling Ma
Summary: This paper proposes a high-speed video adaptive transmission and storage method based on the camera link transmission protocol, which addresses the problems of data overflow and frame loss during transmission and storage. By utilizing adaptive memory buffer pool and adaptive hard disk buffer, the method achieves efficient and stable writing of data. The experimental results demonstrate its broad application value in high-precision vision measurement systems.
Article
Remote Sensing
Zhonghua Hong, Hongyang Zhang, Jinmeng Gao, Xiaohua Tong, Ruyan Zhou, Haiyan Pan, Yun Zhang, Yanling Han, Jing Wang, Shuhu Yang
Summary: This paper proposes a method that applies a combined stereo adjustment model to estimate the collapse height of buildings during earthquakes and thoroughly verifies its feasibility through experiments. By using laser altimetry points and tie points from HRSSIs, the stereoscopic positioning accuracy of HRSSIs is improved. Then, an automatic building corners extraction method is used to assess building damage based on the height change of building corners. The experimental results indicate that the proposed method can accurately assess the degree of building damage.
INTERNATIONAL JOURNAL OF REMOTE SENSING
(2023)
Article
Engineering, Electrical & Electronic
Haiyan Pan, Haoxin Chen, Zhonghua Hong, Xianglei Liu, Runjie Wang, Ruyan Zhou, Yun Zhang, Yanling Han, Jing Wang, Shuhu Yang, Zhenling Ma
Summary: Accurate surface water mapping is crucial for protecting ecosystem environments, but there are few studies on fine extraction of water body boundaries. In this article, a novel boundary enhancement network is proposed to improve the extraction of water boundary information. The network consists of three modules and adopts the Sobel edge loss function. Experimental results show that the proposed network achieves high accuracy in water body extraction.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
