One of the most important tasks in advanced transportation systems is road extraction. Extracting region from high-resolution remote sensing imagery challenging due to complicated background such as buildings, trees shadows, pedestrians and vehicles rural networks that have heterogeneous forms with low interclass high intraclass differences. Recently, deep learning-based techniques presented a notable enhancement image segmentation results, however, them still cannot preserve boundary...

One of the worst environmental catastrophes that endanger Australian community is wildfire. To lessen potential fire threats, it helpful to recognize occurrence patterns and identify susceptibility in wildfire-prone regions. The use machine learning (ML) algorithms acknowledged as one most well-known methods for addressing non-linear issues like wildfire hazards. It has always been difficult analyze these multivariate disasters because modeling can be influenced by a variety sources...

Building objects is one of the principal features that are essential for updating geospatial database. Extracting building from high-resolution imagery automatically and accurately challenging because existence some obstacles in these images, such as shadows, trees, cars. Although deep learning approaches have shown significant improvements results image segmentation recent years, most neural networks still cannot achieve highly accurate with correct map when processing remote sensing...

Urban vegetation mapping is critical in many applications, i.e., preserving biodiversity, maintaining ecological balance, and minimizing the urban heat island effect. It still challenging to extract accurate covers from aerial imagery using traditional classification approaches, because categories have complex spatial structures similar spectral properties. Deep neural networks (DNNs) shown a significant improvement remote sensing image outcomes during last few years. These methods are...

Building extraction with high accuracy using semantic segmentation from high-resolution remotely sensed imagery has a wide range of applications like urban planning, updating geospatial database, and disaster management. However, automatic building non-noisy map obtaining accurate boundary information is big challenge for most the popular deep learning methods due to existence some barriers cars, vegetation cover shadow trees in remote sensing imagery. Thus, we introduce an end-to-end...

Road network extraction from remotely sensed imagery has become a powerful tool for updating geospatial databases, owing to the success of convolutional neural (CNN) based deep learning semantic segmentation techniques combined with high-resolution that modern remote sensing provides. However, most CNN approaches cannot obtain high precision maps rich details when processing imagery. In this study, we propose generative adversarial (GAN)-based approach road aerial part presented GAN...

Road extraction from digital images is of fundamental importance in the context automatic mapping, effective urban planning and updating GIS databases. Very high spatial resolution (VHR) imagery acquired by airborne space borne sensors main source for accurate road extraction. Manual techniques are fading away as they time consuming costly. Hence, method that significantly more automated has become a research hotspot remote sensing information processing. This paper proposes semi-automatic...

In this study, we present a new automatic deep learning-based network named Road Vectorization Network (RoadVecNet), which comprises interlinked UNet networks to simultaneously perform road segmentation and vectorization. Particularly, RoadVecNet contains two networks. The first with powerful representation capability can obtain more coherent satisfactory maps even under complex urban set-up. second is linked the vectorize by utilizing all of previously generated feature maps. We utilize...

Existing automated road extraction approaches concentrate on regional accuracy rather than shape and connectivity quality. Most of these techniques produce discontinuous outputs caused by obstacles, such as shadows, buildings, vehicles. This study proposes a connectivity-preserving identification deep learning-based architecture called SC-RoadDeepNet to overcome the results quality connectivity. The proposed model comprises state-of-the-art network, namely, recurrent residual convolutional...

Terrestrial features extraction, such as roads and buildings from aerial images using an automatic system, has many usages in extensive range of fields, including disaster management, change detection, land cover assessment, urban planning. This task is commonly tough because complex scenes, where road objects are surrounded by shadows, vehicles, trees, etc., which appear heterogeneous forms with lower inter-class higher intra-class contrasts. Moreover, extraction time-consuming expensive to...

Understanding urban dynamics, such as estimating population, development, and several other uses, necessitates up-to-date large-scale building maps. Since aerial imagery provides enough textural structural details, it has been utilized a critical data source for detection. However, accurate mapping of objects from is challenging task. This problem attributed due to presence vegetation shadows in images that present similar spectral values transparency class. To deal with the issues mentioned...

In the present work, a deep learning-based network called LeNet is applied for accurate grassland map production from Sentinel-2 data Greater Sydney region, Australia. First, we apply technique to base date (non-seasonal) make vegetation maps. Then, combine short time-series (seasonal) and enhanced index (EVI) information imagery improve classification results generate high-resolution The proposed model obtained an overall accuracy (OA) of 88.36% mono-temporal data, 92.74% multi-temporal...

A novel hybrid technique for road extraction from UAV imagery is presented in this paper. The suggested analysis begins with image segmentation via Trainable Weka Segmentation. This step uses an immense range of features, such as detectors edge detection, filters texture, noise depletion and a membrane finder. Then, level set method performed on the segmented images to extract features. Next, morphological operators are applied improving precision. Eventually, precision calculated basis...

This paper proposes a model to identify the changing of bare grounds into built-up or developed areas. The is based on fuzzy system and Ordered Weighted Averaging (OWA) methods. proposed consists four main sections, which include physical suitability, accessibility, neighborhood effect, calculation overall suitability. In first two parts, suitability accessibility were obtained by defining inference systems applying required map data associated with each section. However, in order calculate...

Abstract Grass pollen is a globally prevalent allergen, known to trigger allergic reactions such as hay fever and asthma. Australia, in particular, exhibits one of the highest rates asthma prevalence morbidity. Accurate mapping grass sources crucial for enhancing capabilities forecast systems. This especially important urban landscapes, where allergenicity associated with spaces has recently garnered increased attention. However, spatial distribution landscapes not well represented existing...

The primary goal of this research is to see how effective cloud-based computing services such as Google Earth Engine (GEE) platform are at classifying multitemporal satellite images and producing high-quality land cover maps for the target year 2020, with possibility using it on a larger-scale area metropolitan Melbourne test site. To create maps, GEE utilized analyze total 80 Landsat-8 images. support vector machine (SVM) approach used classify Moreover, we use spectral bands, indices,...