kntu
Engineering Journal of Geospatial Information Technology
2008-9635
8
2
2020
9
1
Hyperspectral Images Classification by Combination of Spatial Features Based on Local Surface Fitting and Spectral Features
1
19
FA
Behnam
Asghari Beirami
K.N. Toosi University of Technology
Asghari1370@Gmail.com
Y
Mehdi
Mokhtarzadeh
K.N. Toosi University of Technology
N
10.29252/jgit.8.2.1
Hyperspectral sensors are important tools in monitoring the phenomena of the Earth due to the acquisition of a large number of spectral bands. Hyperspectral image classification is one of the most important fields of hyperspectral data processing, and so far there have been many attempts to increase its accuracy. Spatial features are important due to their ability to increase classification accuracy. In the present paper, a new method is proposed for the spatial features generation of hyperspectral images based on local surface fitting technique. In this method, a surface is fitted to the gray level intensity of the image in the local window around each pixel, and the fitted coefficients, the coefficients of the first and second fundamental forms, curvatures, divergence of the gradient, the area of the gray level intensity of the image and the volume enclosed below the surface are produced in the various window sizes as spatial features. Proposed spatial features stacked with spectral features and form the spectral-spatial vector. this rich spatial-spectral vector is classified with K-nearest neighbor and support vector machine classifiers. The experiments of this paper that are conducted on two real hyperspectral images in agricultural and urban areas show the superiority of the proposed method. The final results show that the overall accuracy of the proposed method in the best case is 7% higher than other competitor methods.
Classification, Hyperspectral Images, Local Surface Fitting Features, Texture, Feature extraction.
http://jgit.kntu.ac.ir/article-1-793-en.html
http://jgit.kntu.ac.ir/article-1-793-en.pdf
kntu
Engineering Journal of Geospatial Information Technology
2008-9635
8
2
2020
9
1
Sensitivity Analysis of Brown Model Waveform in Radar Altimetry
21
38
FA
Reza
Arabsahebi
K.N. Toosi University of Technology
r.arabsahebi@mail.kntu.ac.ir
Y
Behzad
voosoghi
K.N. Toosi University of Technology
N
Mohammad-Javad
Tourian
University of Stuttgart
N
10.29252/jgit.8.2.21
In satellite altimetry (radar altimetry), the altimeter emits a pulse, with known power, to the earth surface and receives it back continuously to determine of the sea surface height. The time series of the mean returned power is recorded individually at satellite as the so-called waveform. Analytical model for the waveform is first introduced by Brown, which consists of six parameters: significant wave height, epoch, skewness of surface, off-nadir angle, thermal noise and backscatter coefficient. The midpoint of the waveform has the main role in determining the distance between sensor and water surface (range). Then, assessing sensitivity of the model to these parameters plays an important role in determining reliability of the range parameter. In this study, the received waveform sensitivity analysis to the mentioned parameters is done for simulated waveform of JASON-2 radar altimetry mission. Then, in order to demonstrate effect of the parameters in observed JASON-2 data, 22 points are chosen over Persian Gulf and Oman Sea from cycle 208 and sensitivity analysis is performed using these points. Our results indicate that the mid-point of the waveform shape is less affected by significant wave height, skewness of surface and off-nadir angle but epoch of half height affects mid-point position significantly, which results in inaccurate range calculation. Because the process of determining the waveform components is a repetitive process, we discuss the choice of initial values of the parameters for using the least squares estimation.
Radar altimetry, sensitive analysis, waveform, Brown model, least square method
http://jgit.kntu.ac.ir/article-1-794-en.html
http://jgit.kntu.ac.ir/article-1-794-en.pdf
kntu
Engineering Journal of Geospatial Information Technology
2008-9635
8
2
2020
9
1
Introducing the improved Forest Canopy density (FCD) model for frequent assessment of Hyrcanian forest
39
58
FA
Masoud
Taefi Feijani
Aerospace Research Institute
m_taefi@ari.ac.ir
Y
Saeed
Azadnejad
Aerospace Research Institute
N
10.29252/jgit.8.2.39
Mapping of forest extent is a prerequisite to acquire quantitative and qualitative information about forests and to formulate management and conservation strategies. forest canopy density (FCD) model is one of the useful RS methods for forest mapping using satellite images. One of the most serious challenges in FCD model is the weakness in the calculation of canopy density in low density forests as well as plain forests. Due to the existence of chlorophyll in croplands, shrubs, pastures, etc., FCD model has difficulty to determinate the forests areas from the other mentioned land cover. Hence, this paper is focused on improving the performance of FCD model to overcome this limitation. This improvement yield by adding a new forest color composite index (FCCI) and removing non-forest vegetation using the average kernel and DEM regard to standard forest definition. In this study, in order to implement and evaluate the performance of the improved model, time series of Landsat images acquired from USGS Landsat standard level-2 products archive. In this study, Landsat time series images acquired from USGS Landsat standard level-2 products were used to estimate forest canopy density in Hyrcanian forests of northern Iran. The results indicated the higher accuracy of the proposed model. Moreover, overall accuracy and kappa index of the model were 10% and 24% superior to initial model, respectively. As a second objective, in order to implement and evaluate the performance of the improved model, canopy changes of the Hyrcanian forests were also examined. In general, the results of this study showed that the total area of Hyrcanian forest increased from 154,272 hectares from 1987 to 2017. Mazandaran, Gilan and Golestan provinces contributed 75,070, 47,615 and 31,567 hectares respectively. In addition, the results showed that the area of Hyrcanian forests decreased by 17,631 hectares between 2009 and 2017.
Hyrcanian forest, Forest canopy density model, Landsat time series data, plain forests.
http://jgit.kntu.ac.ir/article-1-795-en.html
http://jgit.kntu.ac.ir/article-1-795-en.pdf
kntu
Engineering Journal of Geospatial Information Technology
2008-9635
8
2
2020
9
1
Improving the clay, silt and sand of soil prediction by removing the influence of moisture on reflectance using EPO
59
74
FA
Saham
Mirzaei
University of Tehran
N
University of Tehran
ali.darvishi@ut.ac.ir
Y
Hossein Ali
Bahrami
Tarbiat Modares University
N
Seyyed Kazem
Alavipanah
University of Tehran
N
Ali jafar
Mousivand
Tarbiat Modares University
N
10.29252/jgit.8.2.59
Moisture is one of the most important factors that affects soil reflectance spectra. Time and spatial variability of soil moisture leads to reducing the capability of spectroscopy in soil properties estimation. Developing a method that could lessen the effect of moisture on soil properly prediction using spectrometry, is necessary. This paper utilises an external parameter orthogonalisation (EPO) algorithm to remove the effect of soil moisture from spectra for the estimation of texture element contents of soil. ?????The reflectance of 175 soil samples with nominal moisture contents approximately air dry, 6, 12, 18, 24, 30 and 36% were measured. Cross validation was adapted to determine the optimum number of components in the EPO matrix model-coupled. PLSR method has been used for estimation of soil propetrices. The result shows that the presence of moisture leads to reducing the acuraccy reduction of clay (from R2=0.70 to R2=0.38), silt (from R2=0.34 to R2=0.20), and sand (from R2=0.40 to R2=0.30) prediction. Reduction in the accuracy increases by increasing the moisture levels. Removing the effects of moisture from the soil reflectance by EPO algorithm lead to improving 0.23, 0.12 and 0.16 in R2 of clay, silt and sand prediction by PLSR model, respectively. Therefore, using the EPO-PLS method, in any moisture level, VNIR spectroscopy is a viable tool for estimation of soil texture elements.
EPO, PLSR, spectroscopy, soil moisture, soil texture.
http://jgit.kntu.ac.ir/article-1-796-en.html
http://jgit.kntu.ac.ir/article-1-796-en.pdf
kntu
Engineering Journal of Geospatial Information Technology
2008-9635
8
2
2020
9
1
3D Detection of Power-Transmission Lines in Point Clouds Using Random Forest Method
75
91
FA
Mohammad Bagher
Mohammadi moghaddam
University of Tehran
N
Farhad
Samadzadegan
University of Tehran
N
Farzaneh
Dadrass Javan
University of Tehran
fdadrasjavan@ut.ac.ir
Y
10.29252/jgit.8.2.75
Inspection of power transmission lines using classic experts based methods suffers from disadvantages such as highel level of time and money consumption. Advent of UAVs and their application in aerial data gathering help to decrease the time and cost promenantly.
The purpose of this research is to present an efficient automated method for inspection of power transmission lines based on point clouds achieved by aerial data. The proposed method followed by five steps: removing noise on point clouds and filtering point clouds in order to divide it into two parts of ground points and non-ground points, features extraction from non-ground point clouds and finally, power lines classified for 3d detection of power lines. For capability assessment of the proposed method, wo different data sets as aerial RGB based UAV imagery ad aerial laser based data is applied. Accuracy of the proposed method was 97.05% in total classification and 98.80% in power lines detection for dataset 1 taken over an urban area with spectral features. The total accuracy in classification was 95.48% and 96.81% in power lines detection for dataset 2 that taken from a rural area.
Power-transmission lines detection, Random forest Classification, Laser Scanner Point Cloud, UAV based data.
http://jgit.kntu.ac.ir/article-1-797-en.html
http://jgit.kntu.ac.ir/article-1-797-en.pdf
kntu
Engineering Journal of Geospatial Information Technology
2008-9635
8
2
2020
9
1
Development a split window algorithm to estimate land surface temperature from Sentinel -3 satellite data
93
113
FA
Masoud
Heydari
University of Tehran
N
Mehdi
Akhoondzadeh Hanzaei
University of Tehran
makhonz@ut.ac.ir
Y
10.29252/jgit.8.2.93
Land Surface Temperature (LST) is an important indicator of the study of energy balance models at the earth's surface and the interactions between the Earth and the atmosphere on a regional and global scale. To date, different algorithms have been developed in the last few decades to determine the land surface temperature using various satellite images. In this study, a new split window method for estimating the land surface temperature using Sentinel-3A Satellite Radiometer (SLSTR) images is presented. The advantage of the proposed method is the introduction of atmospheric water vapor into the split window algorithm, which plays an important role in estimating the LST. The LST was calculated by the proposed method and three other split window algorithms. Then the results of the proposed method and three split window algorithms were compared with the ASTER, MODIS and Sentinel-3 LST products. The RMSE value of the proposed method for the study area of East Tehran was 3.49, 1.22 and 1.26 K, respectively, compared to the ASTER, MODIS and Sentinel-3 LST products which is lower than the RMSE obtained from other split window algorithms. Also the proposed method and three split window algorithms were implemented for northwest of Isfahan and Kermanshah. For Kermanshah study area results were much better than the other two cases and also other methods which the RMSE of the proposed method was calculated to be 1.05 Kelvin while the RMSE of the other methods was 1.19, 1.28 and 1.56 Kelvin.
Land Surface Temperature (LST), Split-Window Algorithm, Sentinel-3, SLSTR.
http://jgit.kntu.ac.ir/article-1-798-en.html
http://jgit.kntu.ac.ir/article-1-798-en.pdf