kntu
Engineering Journal of Geospatial Information Technology
2008-9635
10
1
2022
6
1
Improving the performance of RPNet with LDA for extracting the deep features for the classification of hyperspectral images
1
15
FA
Behnam,
Asghari Beirami
K. N .Toosi university
Mehdi
Mokhtarzade
K. N .Toosi university
In recent years, deep models have achieved great success in various fields of image processing. These models have been used in some research fields of hyperspectral data processing, such as; classification and target detection. The random patches network (RPNet) has recently been proposed to extract hierarchical deep features for hyperspectral image classification. RPNet is important as it is an unsupervised method, and as a consequence, it has a fast performance to extract deep features. Despite the good performance of this network, due to the usage of the principal component analysis (PCA) method in its main structure, maximum discrimination between classes is not guaranteed in extracted features. Therefore, in this paper,in order to improve the performance of RPNet, a new method called LDA-RPNet based on linear discriminant analysis (LDA) is proposed. Experiments on two hyperspectral datasets, Indian Pines and Pavia University, show that the LDA-RPNet can extract more compact and suitable features for classifying hyperspectral images. Also, based on the experiments, the LDA-RPNet can increase the overall accuracy by up to 2.5% compared to the classical RPNet.
kntu
Engineering Journal of Geospatial Information Technology
2008-9635
10
1
2022
6
1
Super Resolution Mapping Based on Spatial-Spectral Attraction Model and the New Class Allocation Approach
17
37
FA
Mahnaz
Dastjani
Faculty of Geodesy and Geomatics Engineering, K.N.Toosi University of Technology
Mohammad Javad
Valadan Zoej
Faculty of Geodesy and Geomatics Engineering, K.N.Toosi University of Technology
Mojtaba
Jannati
Faculty of Geodesy and Geomatics Engineering, K.N.Toosi University of Technology
The mixed pixels influence the overall accuracy of land cover maps produced by using the remote sensing images with different spatial resolutions. In recent years, soft-then-hard super resolution mapping (STHSRM) has been proposed to solve the problem of mixed pixels. This method estimates soft attribute values for land cover classes at the subpixel scale level and then allocates classes for subpixels according to the soft attribute values. Subpixel/pixel spatial attraction model (SPSAM) calculates soft attribute values for each class at fine pixels by spatial attraction between subpixels and their neighboring pixels. UOC (Units of Class) allocates classes to subpixels in units of class. First, a visiting order for all classes is predetermined. Then, according to the visiting order, the subpixels belonging to the being visited class are determined by comparing the soft attribute values of this class. The remaining subpixels are used for the allocation of the next class. This paper proposed a new spatial-spectral attraction model to estimate the soft attribute value for each class at each subpixel. Also it presents a novel class allocation approach based on UOC technique for STHSRM algorithm. The proposed class allocation approach produces the optimal location of subpixels by defining the cost function and calculating the corresponding cost of spatial arrangement of sub-pixels in different visiting order of classes. The technique is applied to Worldview-3 and ROSIS-03 images. A comparison between the results obtained through the proposed approach and an existing super-resolution mapping technique is introduced. The results show that the proposed algorithm is able to produce higher SRM accuracy than the other approaches especially in linear feature and class boundaries. The improvement value of the adjusted Kappa coefficient of the proposed algorithm related to the spatial attraction model with the AUOC class allocation technique in the scale factor 2, is 0.053 and 0.032, respectively.
kntu
Engineering Journal of Geospatial Information Technology
2008-9635
10
1
2022
6
1
The impact of access to geospatial information and land-use on users' travel behavior in disruption management in road networks of smart city
39
67
FA
Fateme
Mahdavi
University of Tehran
Rahim Ali
Abbaspour
University of Tehran
Mahsa
Naseri
University of Monash
Today, city smartization and access to published information, influence the daily decisions of the citizens and changes their travel behavior. Although this information is an important step towards smart city and can be of great help to people, it can have side effects. In critical network situations, when a route is disrupted by an accident or route repair, spatial information providers offer options that are completely passenger-oriented and don't care about criteria such as spatial features, landuses, and residents's status. In the present paper, first the effect of access to spatial information on passenger behavior is investigated and it is shown that providing passenger-oriented spatial information to passengers may reduce the utility of the other travelers and residents; Therefore, a solution has been proposed that, on the one hand, maximizes the utility of passengers, on the other hand, maintains the utility of other people involved in this issue. For this purpose, in designing spatial information provided to travelers, the criteria of landuses have been taken into consideration. The agent-based model has been used to evaluate the proposed solution and three scenarios of "no information", "advanced information" and "advanced information considering specific landuses" have been used to evaluate passenger behavior. The data used to evaluate the proposed idea relates to the port of Elizabeth in Nelson Mandela Bay. The results show that having access to information can lead to a reduction in total travel time, a reduction in potential passenger delays, and an increase in passenger satisfaction in a network disruption. It has been shown that using special landuses in spatial data designing not only maintains the aforementioned benefits for travelers, but also provides comprehensive benefits for all those involved.
kntu
Engineering Journal of Geospatial Information Technology
2008-9635
10
1
2022
6
1
Comparison Study of Signal Processing Algorithms for 3D SAR Imaging of MM-WAVE GBSAR System
69
87
FA
Benyamin
Hosseiny
Department of Surveying and Geospatial Engineering, College of Engineering, University of Tehran
Jalal
Amini
Department of Surveying and Geospatial Engineering, College of Engineering, University of Tehran
Safieddin
Safavi-Naeini
Faculty of Electrical and Computer Engineering, University of Waterloo, ON., Canada
This paper evaluates and compares the three-dimensional imaging algorithms for an mm-wave ground based synthetic aperture radar system. There has been significant attention to the development of new ground-based synthetic aperture radar (GBSAR) systems by increasing the demands for various radar remote sensing applications and data. GBSAR systems have unique capabilities, including optimum visual angle to the area of interest, high imaging rate, and low manufacturing and maintenance costs. However, the drawbacks of GBSAR systems can be their limited length of synthetic aperture and high variation between the near and far range comparing to the airborne and satéllite systems. These can affect the received signals and, consequently, the final radar image. To this end, in this paper, three signal processing algorithms, including the Backprojection (BP), Fourier Transform (FT), and Range Migration (RMA), are evaluated for three-dimensional SAR imaging of a GBSAR. This system operates in W frequency band and consists of a two-dimensional mechanical rail to generate a planar synthetic aperture. The above algorithm were investigated in a simulation environment using two different experiments, and the results were evaluated with four metrics, including angular resolution, peak sidelobe ratio (PSLR), integrated sidelobe ratio (ISLR), and signal-to-clutter ratio (SCR). According to the obtained results, all three algorithms presented acceptable imaging results. However, RMA demonstrated a high sensitivity of the target reflectivity to its distance from the zero Doppler line. Furthermore, RMA had more stability in decreasing the angular resolution by increasing the target’s range than the BP and FT algorithms. In contrast, BP and FT obtained poor results in near-field areas. In the case of signal compression, generally, RMA got poor results compared to the other two algorithms, which led to inappropriate results in far distances. Because of having a similar attitude, BP and FT, mostly obtained similar results. However, FT obtained more appealing results with better angular resolution, while the BP algorithm demonstrated slightly better signal compression.
kntu
Engineering Journal of Geospatial Information Technology
2008-9635
10
1
2022
6
1
Investigation of infinitesimal and finite distortions criteria in the use of Transverse Mercator and Universal Transverse Mercator map projections (Case study: Iran region)
89
107
FA
Fateme
Esmaeili
University of Isfahan
Hamid
Mehrabi
University of Isfahan
Vahab
Nafisi
University of Isfahan
Selection of a proper criterion to evaluate map projections' distortions has always been one of the important issues in the field of cartography. One of the classifications for distortions criteria is based on their computational scale. On one hand infinitesimal scale criteria estimate angular and areal distortions based on Tissot indicators, on the other hand finite scale criteria assess shape and areal distortions based on region discretization. In this research, random discretization was replaced with fixed discretization in the computational algorithm of finite scale criteria and the performance of distortions criteria of infinitesimal and finite scale in TM systems and UTM in the study of Iran region was investigated. Although fixed elements remove the limitations of random elements like unrepeatable calculations and lower elements' density in border areas, the results of this study indicate that finite scale criteria are independent of the map projections' characteristics. In order to provide a numerical index for comparing distortions of map projections by averaging the numerical values of criteria, these indexes did not work efficiently in determining the distortions of the projections. Therefore, using only one numerical index to map projections' distortions in regional maps is not significant and it is more reliable to study the pattern of numerical changes of infinitesimal scale criteria to evaluate map projections' distortions in regional maps. According to the results of this study, the numerical range of areal distortions of infinitesimal in the TM projection is up to ten times larger than the UTM system.
kntu
Engineering Journal of Geospatial Information Technology
2008-9635
10
1
2022
6
1
Automatic extraction of roadside transmission poles using mobile laser scanner data
109
134
FA
zahra
chamani
Tafresh University, Iran
Hamid
Bagheri
Technical and Vocational University (TVU)
Heidar
Rastiveis
University of Tehran
Nowadys, mobile terrestrial laser scanning systems (MTLS) have made great strides in collecting three-dimensional (3D) data with high speed and accuracy, as well as high point densities from the road environemnts. Because the manual extraction of the powerlines from the MTLS data is time-consuming, costly and laborious, proposing an automated method using a computer agent has become a scientific challenge as it requires less cost and human labour. In this regard, the proximity of other complications to the powerlines, the presence of natural features such as trees, data incompleteness, and the uniform spatial distribution or shape pattern of the road objects are the main challenges of the powerline detección process. The pulposa of this saudí is to automatically detect powerline poles on the roadside from MTLS point clouds. In the proposed method, the ground points are firstly removed from the data using the Simple Morphological Filter (SMRF) algorithm. Then, the DBSCAN clustering algorithm is employed to classify the remaining points based on the local density information of the points. After that, genetic clustering is used to remove the extra points of the powerline considering the density and intensity characteristics of the points. Eventually, the remaining dispersed points are deleted and power light beams are extracted accurately and neatly. The proposed method was tested and evaluated using four sample sections of the data with different complications. The comparison between the extracted powerlines from the proposed algorithm and the manually extracted results showed the high ability in extracting the powerlines from the MTLS points clouds so that the number of the extracted powerlines is the same.