Figure 6 - uploaded by Siniša Drobnjak
Content may be subject to copyright.
Source publication
In this study, land cover types and degree of urbanization in Belgrade test area were analyzed on the basis of the classification results acquired using object-oriented image analysis approaches. LANDSAT 7 with 7 spectral bands was used to carry out the image classification and ground truth data were collected from the available maps and personal k...
Similar publications
In this study, land cover types and degree of urbanization in Belgrade test area were analyzed on the basis of the classification results acquired using object-oriented image analysis approaches. LANDSAT 7 with 7 spectral bands was used to carry out the image classification and ground truth data were collected from the available maps and personal k...
The city of Safranbolu in Turkey contains well-preserved items of many civilizations (Roman, Byzantine, Seljuk and Ottoman). Although the city has been well preserved in comparison to other heritage sites and studies have been done by different groups at different times, there is a lack of an integrating study covering all aspects of the cultural h...
Citations
... Satellite images are useful equipment for forest monitoring, and remote sensing research has become a very effective method. Satellite images can be used to explore the borders between different types of vegetation, the degree of vegetation development, vegetation morphology, forest health, tree canopy humidity, diverse textures, biomass, and a variety of other parameters (Drobnjak et al., 2013;Bakrač et al., 2018;Drobnjak et al., 2018). ...
The objective of this research is to report results from a new ensemble method for vegetation classification that uses deep learning (DL) and machine learning (ML) techniques. Deep learning and machine learning architectures have recently been used in methods for vegetation classification, proving their efficacy in several scientific investigations. However, some limitations have been highlighted in the literature, such as insufficient model variance and restricted generalization capabilities. Ensemble DL and ML models has often been recommended as a feasible method to overcome these constraints. A considerable increase in classification accuracy for vegetation classification was achieved by growing an ensemble of decision trees and allowing them to vote for the most popular class. An ensemble DL and ML architecture is presented in this study to increase the prediction capability of individual DL and ML models. Three DL and ML models, namely Convolutional Neural Network (CNN), Random Forest (RF), and biased Support vector machine (B-SVM), are used to classify vegetation in the Eastern part of Serbia, together with their ensemble form (CNN-RF-BSVM). The suggested DL and ML ensemble architecture achieved the best modeling results with overall accuracy values (0.93), followed by CNN (0.90), RF (0.91), and B-SVM (0.88). The results showed that the suggested ensemble model outperformed the DL and ML models in terms of overall accuracy by up to 5%, which was validated by the Wilcoxon signed-rank test. According to this research, RF classifiers require fewer and easier-to-define user-defined parameters than B-SVMs and CNN methods. According to overall accuracy analysis, the proposed ensemble technique CNN-RF-BSVM also significantly improved classification accuracy (by 4%).
... The object-oriented method was selected for image classification over the entire study area [40][41][42]. Object-oriented classification operates on image objects, rather than a single pixel. Image objects refer to homogeneous, spatially contiguous regions obtained by dividing an image. ...
Investigation of urban expansion can provide a better understanding of the urbanization process and its driving forces, which is critical for environmental management and land use planning. Total of 514 sampling points from the aerial photos and field sampling were applied to assess the image accuracy. A Conversion of Land Use and its Effect at Small Region Extent (CLUE-S) model was established to simulate the urbanization process at the township level in the North Xinjiang Economic Zone (NXEZ) of western China. Historical land use and land cover changes with multi-temporal remote sensing data were retrieved, and the underlying driving forces were explored by training the CLUE-S model. Moreover, future changes in urban development were simulated under different scenarios. Results showed that the overall accuracy reaches larger than 80% for the years of 2002, 2005, and 2007, and the corresponding kappa coefficient is bigger than 0.8. The NXEZ is at a premature development stage compared with urban clusters in eastern China. Before 1999, the driving force in this region was primary industry development. In recent years, secondary industries started to show significance in urbanization. These findings indicate that the industrial base and economic development in the NXEZ are still relatively weak and have not taken a strong leading role. When industry and population become the main driving factors, the regional economy will enter a new stage of leap-forward development, which in turn will stimulate a new round of rapid urbanization.
... where NIR and RED values are the infrared and red bands of the electromagnetic spectrum, respectively. In this research, the NDVI map was generated using eCognition software (Drobnjak,Ćirović, Sekulović, & Regodić, 2013). Human activities and vehicular movement on roads provide suitable opportunities for negligent and accidental manmade forest fires. ...
... Also, satellite research has some advantages in studying vegetation over conventional methods, among which are [1]: Only radiometric, spatial and spectral enhanced images are ready for further digital analysis in order to obtain the desired data for the purposes of vegetation classification. Grouping of pixels into thematic categories, classes using statistical methods by determining the correlation between their digital values is called classification and represents one of the most demanding operations in computer processing of images in terms of operator knowledge [2]. Classification methods practically involve analyzing the content of the image and combining pixels into the appropriate data categories. ...
... This is achieved through the statistical grouping of pixels into thematic categories based on their digital values, the relationship of the contents of the entities, known as the "class". The classification result represents the separation of one or more classes depending on the needs of the research [1,2]. ...
... One of the privileges of ANN method in comparison to traditional statistical methods is that the networks are free in distribution i.e. the training and recalling are dependent on the linear combination between data patterns and are independent of input data [16]. However, the reasons for the success of ANN in classification can be summarized as: (1) there is no need for pre-assumption in data distribution, (2) it permits the user to make use of the initial knowledge regarding classes and possible limitation, (3) the method allows management of the spatial data from multiple sources and can achieve their classification results equally [12]. ...
This article is connected with methods for automatic vegetation extraction using combination of multispectral satellite images and aerial photogrammetric images. Main goal of this paper is to suggest an optimal solution for easy mapping and updating vegetation areas on topographic maps. Vegetation areas are extracted on two ways: by visual interpretation and mapping using machine learning classification methods of satellite images and aerial photogrammetric images. Comparison of the results between these two approaches showed large geometric similarity. In this study, vegetation areas were extracted on the basis of the classification results acquired by machine learning algoritms approaches. Combination of SENTINEL-2A with 12 spectral bands satellite images and aerial photogrammetric images were used to carry out the image classification and ground sample data were collected from the available maps and personal knowledge. Machine learning classification algorithms were evaluated through ERDAS Imagine and R software. During the implementation, several different machine learning algorithms were tested for image segmentation and classification such as: Support Vector Machine-SVM, Random Forest-RF and Artifacial Neural Network-ANN. Results obtained from vegetation classification process showed that the Random Forest approach gave the most accurate results (including higher producer`s and user`s accuracy for most of the vegetation classes) for vegetation extraction using satellite and aerial photogrammetric images. Average classification accuracies were typically around 90-95%.
The primary objective of this conference article is to review the technological procedure of gathering vegetation cover from aerial photogrammetric and satellite images and obtaining information in digital form in order to model digital topographic maps and augment the database of spatial data on vegetation cover (the areas of the map that change the fastest and most dramatically).
This paper presents technical and technological characteristics of pontoon bridges that are in operational use of
the Serbian Armed Forces, Armed Forces of the United States and the Russian Federation. A comparative analysis of the
technical and technological characteristics and pontoon bridges provides an overview of the state of pontoon bridges in the
Army of Serbia, but we also come to a conclusion about the desirable characteristics sets of pontoon bridges that the
engineering units of the Serbian Armed Forces should have. For the purposes of the comparative analysis, the pontoon
bridges, which are the most exploited in the armed forces of the said countries, were used.
This paper shows the basic characteristics of satellite images of remote sensing in managing emergency situations, as well as their ability to use in managing emergency plans. The satellite systems Landsat 8, Sentinel 2A and Modis were analyzed. Analyzing the accuracy of the classification of satellite images in crisis situations, it has been shown that the obtained results for Landsat 8 and Sentinel 2 satellite images show the same level of accuracy. The slight advantage of Landsat 8 is in the detection of water areas and built-up areas, while Sentinel 2 has the advantage of detecting agricultural and forest areas. On the other hand, Modis recordings are characterized by low spatial resolution, so their use in detecting changes in crisis management is significantly reduced compared to Sentinel and Landsat. However, high time resolution enabled the use of these images in different areas, as well as in crisis management situations.
