BoVSG: bag of visual SubGraphs for remote sensing scene classification

Amiri, Khitem
Farah, Mohamed
Leloğlu, Uğur Murat
Remote sensing scene classification is gaining much more interest in the recent few years for many strategic fields such as security, land cover and land use monitoring. Several methods have been proposed in the literature and they can be divided into three main classes based on the features used: handcrafted features, features obtained by unsupervised learning and those obtained from deep learning. Handcrafted features are generally time consuming and suboptimal. Unsupervised learning based features which have been proposed later gave better results but their performances are still limited because they mainly rely on shallow networks and are not able to extract powerful features. Deep learning based features are recently investigated and gave interesting results. But, they cannot be usually used because of the scarcity of labelled remote sensing images and are also computationally expensive. Most importantly, whatever kind of feature is used, the neighbourhood information of them is ignored. In this paper, we propose a novel remote sensing scene representation and classification approach called Bag of Visual SubGraphs (BoVSG). First, each image is segmented into superpixels in order to summarize the image content while retaining relevant information. Then, the superpixels from all images are clustered according to their colour and texture features and a random label is assigned to each cluster that probably corresponds to some material or land cover type. Thus superpixels belonging to the same cluster have the same label. Afterwards, each image is modelled with a graph where nodes correspond to labelled superpixels and edges model spatial neighbourhoods. Finally, each image is represented by a histogram of the most frequent subgraphs corresponding to land cover adjacency patterns. This way, local spatial relations between the nodes are also taken into account. Resultant feature vectors are classified using standard classification algorithms. The proposed approach is tested on three popular datasets and its performance outperforms state-of-the-art methods, including deep learning methods. Besides its accuracy, the proposed approach is computationally much less expensive than deep learning methods.


End-to-end networks for detection and tracking of micro unmanned aerial vehicles
Aker, Cemal; Kalkan, Sinan; Department of Computer Engineering (2018)
As the number of micro unmanned aerial vehicles (mUAV) increases, several threats arise. Hence, there is a need for a system that can detect and track them. In this thesis, an object detection model based on convolutional neural networks for mUAV detection, and a novel end-to-end object tracking architecture are proposed. To solve the scarce data problem for training the detection network, an algorithm for creating an extensive artificial dataset by combining background-subtracted real images is proposed. I...
IPBM: an energy efficient reliable interference-aware periodic broadcast messaging protocol for MANETs
ÜNLÜ, BERK; Ozceylan, Baver; Baykal, Buyurman (Springer Science and Business Media LLC, 2019-07-01)
Mobile ad-hoc networks (MANETs) have been widely employed in many fields including critical information delivery in open terrains as in tactical area, vehicular or disaster area network scenarios. To provide effective network maintenance for those MANETs, it is essential to adopt proper control communication methods, which provide reliable delivery of network information. However, it is difficult to provide control communication that meets the quality of service requirements due to the broadcasting of contr...
Optimized Unmanned Aerial Vehicles Deployment for Static and Mobile Targets' Monitoring
Al-Turjman, Fadi; Zahmatkesh, Hadi; Al-Oqily, Ibrhaim; Daboul, Reda (Elsevier BV, 2020-01-01)
In the recent decade, drones or Unmanned Aerial Vehicles (UAVs) are getting increasing attention by both industry and academia. Due to the support of advanced technologies, they might be soon an integral part of any smart-cities related project. In this paper, we propose a cost-effective framework related to the optimal placement of drones in order to monitor a set of static and/or dynamic targets in the IoT era. The main objective of this study is to minimize the total number of drones required to monitor ...
SWIR objective design using seidel aberration theory
Aslan, Serhat Hasan; Yerli, Sinan Kaan; Keskin, Onur; Department of Physics (2016)
Optical systems are used for increasing the situational awareness and Intelligence, Surveillance and Reconnaissance (ISR) capabilities for military purposes. MWIR (Midwave infrared) and LWIR (Long wave infrared) waveband informations are the first two wavebands information in the atmospheric transmission window that are harnessed in military night vision optical systems. Another candidate of these operable wavebands is the SWIR (Shortwave infrared). Shortwave infrared (SWIR) imaging is an extension of Near ...
Comparison of the detection performance of an FMCW coastal surveillance radar for v and H polarizations
Secmen, M.; Demir, Şimşek; Hizal, A.; Candan, N. (2006-12-01)
The present work supports the use of V-polarization instead of H-polarization under CFAR detection in FMCW coastal surveillance radars due to its statistical advantage. The sea clutter reflectivity, σ0, is about 3-5 dB less for H-polarization than for V-polarization. However, the sea clutter echoes are spikier for H-polarization than V-polarization for a high resolution radar. It has been suggested that for a high resolution radar V-polarization is to be preferred over H-polarization for a fixed threshold d...
Citation Formats
K. Amiri, M. Farah, and U. M. Leloğlu, “BoVSG: bag of visual SubGraphs for remote sensing scene classification,” INTERNATIONAL JOURNAL OF REMOTE SENSING, pp. 1986–2003, 2019, Accessed: 00, 2020. [Online]. Available: