A5061612965- Remote Sensing and LiDAR Applications
- Automated Road and Building Extraction
- Remote-Sensing Image Classification
- Remote Sensing in Agriculture
- 3D Surveying and Cultural Heritage
- Synthetic Aperture Radar (SAR) Applications and Techniques
- Air Quality and Health Impacts
- Wood and Agarwood Research
- Geophysical Methods and Applications
- Statistical and numerical algorithms
- Land Use and Ecosystem Services
- Remote Sensing and Land Use
- Video Surveillance and Tracking Methods
- Rock Mechanics and Modeling
- Soil Geostatistics and Mapping
- Impact of Light on Environment and Health
- Coal Properties and Utilization
- Robotics and Sensor-Based Localization
- Aeolian processes and effects
- Domain Adaptation and Few-Shot Learning
- Advanced Image Processing Techniques
- COVID-19 impact on air quality
- Hydrocarbon exploration and reservoir analysis
- Atmospheric chemistry and aerosols
- Advanced Image and Video Retrieval Techniques
University of Waterloo
2020-2023
University of Tehran
2019-2020
Geospatial Research (United Kingdom)
2019
Geomatics (Norway)
2019
Kharazmi University
2019
Since the shorelines are important geographical boundaries, monitoring shoreline change plays an role in integrated coastal management. With evolution of remote sensing technology, many studies have used optical images to measure and extract shoreline. However, some factors limit use imaging on mapping. Considering that airborne LiDAR data can provide more accurate topographical information, there been investigated using map shorelines. a literature review combines with measurement...
Mobile Laser Scanning (MLS) system can provide high-density and accurate 3D point clouds that enable rapid pavement crack detection for road maintenance tasks. Supervised learning-based algorithms have been proved pretty effective handling such a large amount of inhomogeneous unstructured clouds. However, these often rely on lot annotated data, which is labor-intensive time-consuming. This paper presents semi-supervised point-level approach to overcome this challenge. We propose graph-widen...
With the fast development of 3D data acquisition techniques, topographic point clouds have become easier to acquire and promoted many geospatial applications. Ground filtering (GF), as one most fundamental challenging tasks for post-processing large-scale clouds, has been extensively studied but yet be well solved. To reveal future superior solutions, a comprehensive investigation up-to-date GF studies is essential. However, existing surveys are scarce fail capture latest progress...
Algae blooms have been a serious problem in coastal and inland water bodies across Canada. The temporal spatial variability of algae makes it difficult to use situ monitoring the lakes. This study aimed evaluate potential Sentinel-3 Ocean Land Color Instrument (OLCI) Sentinel-2 Multispectral (MSI) data for algal Lake Erie. Chlorophyll-a (Chl-a)-related products these sensors were tested by using Great Lakes Chl-a NOAA’s over summer 2016 2017, respectively. Our results show that while...
This study presents a novel workflow for automated Digital Terrain Model (DTM) extraction from Airborne LiDAR point clouds based on convolutional neural network (CNN), considering transfer learning approach. The consists of three parts: feature image generation, using ResNet, and interpolation. First, each is transformed into featured its elevation differences with neighboring points. Then, the images are classified ground non-ground ImageNet pretrained ResNet models. points extracted by...
(2021). A Comparative Study of Deep Learning Approaches to Rooftop Detection in Aerial Images. Canadian Journal Remote Sensing: Vol. 47, Large-Scale for Sensor-Driven Mapping, pp. 413-431.
Accurately mapping individual tree species in densely forested environments is crucial to forest inventory. When considering only RGB images, this a challenging task for many automatic photogrammetry processes. The main reason that the spectral similarity between scenes, which can be hindrance most methods. This paper presents deep learning-based approach detect an important multi-use of palm trees (Mauritia flexuosa; i.e., Buriti) on aerial imagery. In South-America, essential indigenous...
Land subsidence is one of the most hazardous phenomena. Rapid rate corresponds to human activities such as subsurface withdrawal. Therefore, local changes in surface elevation and its associated response has potential damage industrial structures. In this research, small baseline subset (SBAS) algorithm applied perform an Interferometric Synthetic Aperture Radar (InSAR) time series analysis within 2003 2010 over a giant oilfield located Zagros fold-thrust belt southwest Iran. Thirty-one...
The Halton Region, as part of the Greater Toronto Area (GTA), is regarded one fastest growing regions in Canada, generating 20% national gross domestic product. It also most desirable places for living and thriving businesses. This research attempts to assess urban expansion Ontario, Canada from 1989 2019 using satellite images, analysis approaches, landscape metrics. Multitemporal Landsat images supervised learning algorithms GIS software were used explore dynamic changes classify nonurban...
Surface displacements associated with the average subsidence due to hydrocarbon exploitation in southwest of Iran which has a long history oil production, can lead significant damages surface and subsurface structures, requires serious consideration. In this study, Small BAseline Subset (SBAS) approach, is multitemporal Interferometric Synthetic Aperture Radar (InSAR) algorithm was employed resolve ground deformation Marun region, Iran. A total 22 interferograms were generated using 10...
This paper explores the effect of COVID-19 outbreaks on human activity through nighttime light images Greater Toronto Area (GTA), Canada. The methods used in this include image preprocessing, classification, and spatial analysis. By using radiance data from VIIRS/NPP products cases comparing pre-pandemic year, impact was analyzed. result shows that during pandemic year monthly average has decreased about 4.3-5.0% compared to year. classification results percentage changes residential areas,...
A constrained extended Kalman filter (CEKF) based on least-squares variance component estimation (LS-VCE) is generally developed by condition equations since the proper prediction of dispersion matrices one main bottlenecks in KF algorithms. Here we investigate four problems which have not been simultaneously considered yet. These are examination non-linearty dynamic model, VCE, general non-linear state constraints and fairly stochastic model. Although a few contributions proposed some...
Building rooftop data are of importance in several urban applications and natural disaster management. In contrast to traditional surveying mapping, by using high spatial resolution aerial images, deep learning-based building rooftops extraction methods efficient accurate. Although more training is preferred tasks, the effect volume on models underexplored. Therefore, paper explores impact performance from very-high-spatial-resolution (VHSR) images methods. To do so, we manually labelled...
Abstract. Automated building footprints extraction from High Spatial Resolution (HSR) remote sensing images plays important roles in urban planning and management, hazard disease control. However, HSR are not always available practice. In these cases, super-resolution, especially deep learning (DL)-based methods, can provide higher spatial resolution given lower images. a variety of applications, DL based super-resolution methods widely used. there few studies focusing on the impact DL-based...
Abstract. The COVID-19 was first declared by World Health Organization (WHO) as global pandemic on March 11th 2020. While most of COVID-related studies have focused epidemiological perspective, the spatial analysis disease outbreak is also important to provide perceptions transmission rates. Therefore, this paper attempts identify potential factors contributing incidence rate at provincial-level in Canada. Three statistical regression models, ordinary least squares (OLS), error model, and...
Abstract Accurately mapping individual tree species in densely forested environments is crucial to forest inventory. When considering only RGB images, this a challenging task for many automatic photogrammetry processes. The main reason that the spectral similarity between scenes, which can be hindrance most methods. This paper presents deep learning-based approach detect an important multi-use of palm trees (Mauritia flexuosa; i.e., Buriti) on aerial imagery. In South-America, essential...
We aim to improve the efficiency of traditional deep learning methods for remote sensing by reducing reliance on annotated data and minimizing training time. Instead using large-scale unimodal image datasets pre-training, we propose use multimodal (text-image pairs), which believe be more effective. To enhance model's generalization performance in domain achieve accurate scene classification, employ Feature Adaptive Embedding Module. For this purpose, introduce a cross-modal comparison...
With the development of social economy, wind turbines are taking up a larger and share new energy sources. The detection number spatial distribution in remotely sensed images holds great scientific significance. Wind difficult to identify remote sensing therefore, fast method based on deep learning is proposed. First, we extract potential turbine candidate regions from speed, slope, land use data. Second, YOLO v5 model was trained using our labeled dataset. Finally, were used for optimal...
LiDAR sensors generate noisy, sparse, and non-uniform point clouds. Although 3D sensing has had significant progress in recent years, it is still expensive to produce high-quality Therefore, given a low-density dataset, cloud upsampling required obtain clouds with high density more complete details. This study addresses an network called PU-GCN on dataset collected indoor environment such as underground parking lot dense uniform dataset. Then, the VoteNet model used for evaluating impact of...
Three constrained extended Kalman filters (CEKF) are developed by making use of condition equations which allows one to predict directly the residuals all variables. The first is a general CEKF algorithm in it supposed that observation equations, system and constraints dynamic problem non-linear functions. Although filter was already investigated few contributions, they assumed some restrictive conditions such as linearity and/or equations. Moreover, this generalization helps deal with...
Landslide is one of the disasters, which occurs throughout world under all climatic conditions and terrain.This leads to loss lives property damage natural environment.Therefore, a need arises for landslide hazard zonation identification potential areas.This research an attempt towards modeling using GIS remote sensing techniques.Phuentsholing-Pasakha highway has been selected study due its importance in economy country.Parameters causing landslides were determined through thorough...
Abstract. British Columba, Canada experienced its record-breaking wildfire season in 2017. The smoke is one of the main sources fine particles with diameters smaller than 2.5 μm (PM2.5). rising level PM2.5 concentrations during would considerable increase risk premature death, especially for people weak immune systems. In this study, satellite optical data collected from 3 km MODIS aerosol depth (AOD) products were adopted to estimate concentration levels derived wildfires Columbia, July...