Article
Geochemistry & Geophysics
Yongfei Li, Dongfang Yang, Shicheng Wang, Hao He, Jiaxing Hu, Huaping Liu
Summary: This article proposes a road-network-based geolocalization method that successfully localizes over whole city areas. It treats the road network matching problem as a point cloud registration problem and introduces global projective-invariant features for solving it. Experimental results demonstrate that the method can accurately and quickly localize aerial images over large areas.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Computer Science, Hardware & Architecture
Wei Wu, Yong Xian, Shaopeng Li, Juan Su, Daqiao Zhang
Summary: In this article, a deep learning framework is proposed for the alignment between aerial image and road based geo-parcel data. The framework includes a pre-processing step of image segmentation, followed by multi-scale deep feature extraction, and a multi-level alignment network. Synthetic image datasets are used to test and verify the performance of the proposed method, which shows effective alignment of image pairs and improvement in matching performance compared to existing methods.
Article
Computer Science, Information Systems
Tanmay Kumar Behera, Sambit Bakshi, Pankaj Kumar Sa, Michele Nappi, Aniello Castiglione, Pandi Vijayakumar, Brij Bhooshan Gupta
Summary: Recent years have seen significant advancements in small-scale remote sensors such as UAVs, particularly in the field of computer-vision tasks like aerial image segmentation. This paper introduces the NITRDrone dataset, which focuses on extracting road networks from aerial images captured at different locations on the NITR campus. Extensive experiments have been conducted to validate the dataset's effectiveness, with U-Net achieving the best performance. The availability of the NITRDrone dataset is expected to boost research and development in visual analysis of UAV platforms.
JOURNAL OF SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY
(2023)
Article
Computer Science, Information Systems
Dongqing Zhang, Yucheng Dong, Zhaoxia Guo
Summary: The paper introduces a novel turning point-based offline map matching algorithm, which improves matching accuracy and efficiency by segmenting the entire trajectory into sub-trajectories and selecting the best-matched path from the K-shortest paths. Extensive experiments show that the algorithm outperforms five benchmark algorithms in terms of correctly matched percentages, incorrectly matched percentages, and matching speeds.
INFORMATION SCIENCES
(2021)
Article
Remote Sensing
Zhu Mao, Xianfeng Huang, Wenyuan Niu, Xuan Wang, Zepeng Hou, Fan Zhang
Summary: This research proposes an instance segmentation method to automatically extract slender urban road facilities (SURFs) from oblique aerial imagery. The method enhances both the predicted bounding box and the segmented binary mask. The proposed method performs superiority in SURF instance segmentation, obtaining an mAP of 0.888 for the predicted bounding box and an mAP of 0.876 for the segmented binary mask.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)
Article
Geochemistry & Geophysics
Ying Fu, Shuaizhe Liang, Dongdong Chen, Zhanlong Chen
Summary: This article proposes an end-to-end online map generation method that combines discrimination and creativity by utilizing a semantic segmentation module and a creative module to mimic human behavior for generating accurate and visually appealing online maps. Extensive experiments with a large dataset consisting of aerial images and online maps from nine regions of six continents demonstrate the superiority of the new design over baseline methods.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Computer Science, Artificial Intelligence
Guanke Chen, Kun Hao, Beibei Wang, Zhisheng Li, Xiaofang Zhao
Summary: This paper proposes an improved power line segmentation model based on Deeplabv3+ (PL-Deeplab), which addresses the problems of complex background interference and thin structure of power lines in aerial images. The model utilizes a multibranch concatenation network (MCNet), a one-shot aggregation feature pyramid (OSAFP), and a feature fusion module (FFM) to accurately segment the power lines. Experimental results on a public dataset demonstrate the superiority of the proposed model in completing the power line inspection task and ensuring security for UAV power line inspection.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Environmental Sciences
Osmar Luiz Ferreira de Carvalho, Osmar Abilio de Carvalho Junior, Cristiano Rosa e Silva, Anesmar Olino de Albuquerque, Nickolas Castro Santana, Dibio Leandro Borges, Roberto Arnaldo Trancoso Gomes, Renato Fontes Guimaraes
Summary: Panoptic segmentation has great potential in remotely sensed data as it combines instance and semantic predictions to detect countable objects and different backgrounds simultaneously. However, challenges such as labeling large images, generating DL samples in the panoptic segmentation format, handling large remote sensing images, and software compatibility issues have hindered the growth of this task. This study addresses these challenges by providing a pipeline for generating panoptic segmentation datasets, software for creating deep learning samples in the COCO annotation format, a novel dataset, compatibility with remote sensing data using Detectron2 software, and evaluation on the urban setting.
Article
Computer Science, Artificial Intelligence
Doruk Oner, Mateusz Kozinski, Leonardo Citraro, Nathan C. Dadap, Alexandra G. Konings, Pascal Fua
Summary: The paper proposes a novel connectivity-oriented loss function for training deep convolutional networks to reconstruct network-like structures, such as roads and irrigation canals, from aerial images. The loss function aims to express the connectivity of roads or canals in terms of disconnections, and prevents unwanted connections between background regions by penalizing unwarranted disconnections.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Computer Science, Software Engineering
Renjie Chen, Craig Gotsman
Summary: In the age of real-time online traffic information and GPS-enabled devices, fast path computation in large road networks is critical, and the A* algorithm is an efficient way to reduce the number of vertices traversed, providing better estimates of minimal cost than other heuristics.
COMPUTATIONAL VISUAL MEDIA
(2021)
Article
Engineering, Electrical & Electronic
Ahmad Almadhor
Summary: This paper introduces a deep learning-based algorithm that modifies the VGG-16 network to enhance image segmentation results by combining features of various CNN networks. The use of dilated convolution kernels, feature pyramids, and channel-wise attention aids in recovering lost features and learning global features for foreground object recovery. The algorithm is tested on two public datasets against top-ranked image segmentation methods.
IEEE SENSORS JOURNAL
(2021)
Article
Remote Sensing
Nicholas Paul Sebasco, Hakki Erhan Sevil
Summary: This research proposes a novel method for identifying and extracting roads from aerial images taken after a disaster. The method uses graph-based image segmentation and includes modifications and improvements to the Efficient Graph-Based Image Segmentation. The proposed method achieves high performance in road extraction and outperforms a similar technique using K-means clustering.
Article
Computer Science, Information Systems
Wenbin Zhu, Hong Gu, Zhenhong Fan, Xiaochun Zhu
Summary: This paper presents a novel method for road target segmentation in autonomous driving based on stereo disparity maps. The proposed method addresses the challenge of selecting appropriate thresholds by using topological persistence threshold analysis. Experimental validation demonstrates the effectiveness and superior performance of the method, suggesting potential for application in autonomous driving systems.
Article
Geochemistry & Geophysics
Zhenhua Xu, Yuxuan Liu, Lu Gan, Yuxiang Sun, Xinyu Wu, Ming Liu, Lujia Wang
Summary: This article proposes a novel approach based on transformer and imitation learning to generate road network graphs vertex-by-vertex using high-resolution aerial images. Comparative experiments demonstrate the superiority of the proposed approach.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Computer Science, Artificial Intelligence
Md. Imteaz Ahmed, Md. Foysal, Manisha Das Chaity, A. B. M. Aowlad Hossain
Summary: In this study, a practical approach for road segmentation is assessed, and a robust model named DeepRoadNet is proposed, which utilizes a pre-trained EfficientNetB7 architecture and residual blocks for road segmentation, achieving better results compared to existing models.
IET IMAGE PROCESSING
(2023)
