جستجوی مقالات مرتبط با کلیدواژه « classification » در نشریات گروه « عمران »

Texture quantization is a useful method for extracting spatial relevance between pixels, which is used in the human brain for image interpretation. Aside from spectral bands, textural features of high spatial resolution image can be used to improve classification accuracy. Finding proper textural features among available features is important for special case studies.

In this paper, two methods based on genetic algorithm (GA) are introduced to choose efficient features. The first is binary GA, which improves classification accuracies through selecting the best textural features. The second one is GA with a variable number of selected features in a refined and full feature space. Results show that the best combination does not necessarily consist of features with improved individual accuracy.

The proposed methods have better accuracy, less number of features, and less computational time when comparing with the simple GA. They could be used based on the number of spectral bands, number of generated features, and train and check pixel number. Second method needs more prerequisite time and could be used for images with fewer bands, train and check pixels, and generated features, because increasing these items increase computational time very much. Second method could be used in large images with more train and check pixels but led to more selected features.

Results obtained on three datasets indicate 7.7 to 50.48 percent improvement in mean accuracy.

The issue of urbanization and monitoring of urban expansion and land use changes using satellite images has become a basic focus in the society. Easy and stable access to satellite data has made it possible to monitor and monitor land changes more accurately; But for optimal use of these images, it is necessary to collect samples of images and then classify their pixels based on regional features and characteristics. This process faces challenges such as data dispersion, which can be solved by using appropriate classification methods. In this study, in order to evaluate the area of land uses in cities, various methods of machine learning have been used. Instead of using a fixed and absolute method for classifying pixels, four different machine learning methods are investigated separately for each image. These diverse methods of machine learning provide the possibility of choosing the best and most efficient method for each image, thus improving the ability to detect and classify pixels for land use areas in cities and increasing accuracy and efficiency.

In this research, the Landsat 9 satellite image has been used to study and analyze different areas of Tehran in 2023. First, the desired image was subjected to the necessary corrections and then four appropriate machine learning algorithms (which included K-nearest neighbor, support vector machine, random forest and maximum likelihood) were used to classify Landsat 9 satellite images related to four different areas of Tehran (including 2, 5 , 21, 22) were used. To evaluate the accuracy of the results, more than 200 check points were created on the image using the Stratified Random method, and then Google Earth Pro was used to check the check points. The overall classification accuracy and kappa coefficient were evaluated as evaluation criteria for the best classification method of image pixels. In the next step, the studied area was divided into equal blocks in order to better understand the area of land uses in that area. Then, using Zonal Statistics, the amount of land use area in each block was investigated.

Based on the methods used, the performance of the SVM method in this study achieved the highest possible accuracy, which is equal to 95%, and the Kappa coefficient, which is 89%. These results may be justified due to the non-uniformity of pixel areas in dense urban environments. In addition, different areas of land, including green areas with an area of 12 square kilometers, barren lands with an area of 64 square kilometers, and built-up areas with an area of 137 square kilometers were also examined in this analysis.

Through this approach, we have presented a highly accurate classification method for the analysis of satellite images related to the Landsat 9 satellite. This method enables a more accurate assessment of the area of land uses and provides urban decision makers and policy makers with a direct link with valuable insights for sustainable development in cities. This can play an effective role in the process of facilitating development plans to improve cities and citizens' lives, because it provides accurate and reliable information that helps strategic decisions in the field of urban development and enables more effective and targeted changes in urban policies and programs.

Identifying the effective parameters on the increase of the accidents severity in the two-lane highway and also the spatial analysis of the accidents occurring in them, could lead to the reduction of the road accidents of this road. Based on this, the present research in addition to identifying section with high crashes, it also identifies the factors affecting the severity of accidents.

In recent years, road accidents have become one of the leading causes of death in Iran. Due to the role of various factors in the occurrence of a road accident, the management of the occurrence of this phenomenon has been associated with many challenges.Due to the presence of several qualitative and quantitative factors in the accident analysis process, this research uses a developed approach based on the theory of Rough Sets to identify the pattern of road accident incidence. By using the proposed model, it will be possible to analyze the occurrence of accidents in multiple dimensions and finally rule out an accident by combining the factors involved. For this purpose, a comprehensive range of factors based on the literature of the subject and the opinions of experts related to the vehicle, driver and the intended environment are discussed. Key findings of the accident management research suggest that the existence of scientific models of accident factors can reduce the number of road accidents. The causes of the accident with the highest number of repetitions were extracted in the form of rules.  The results show that the type of accident depending on the type of vehicle and the weather conditions can have different severity of the accident. In addition, the cause of an accident under various conditions, such as atmospheric conditions, can have a specific incident severity. Thus, the theory of Rough sets performs an acceptable function in analyzing the role of accident factors and its quality in order to respond appropriately after occurrence and prevention.

Pavement inspection is one of the most important steps in implementation of pavement management system and extend efforts have been done to increase the efficiency of this system by using the new technologies. In recent years, transportation agencies focus on creating automatic and more efficient systems for pavement inspection and a large number of researches have been done for this aim. According to progress of computer science, data mining and machine learning as computer-based methods are used more in various areas (such as engineering, medical and economy) and significant results have been achieved. In pavement management area, several researches have been performed with aim of the applying the machine learning especially in pavement distresses evaluation. In this paper, the theoretical concepts have been explained and several models have been created based on deep convolutional networks using transfer learning to detect and classify pavement cracks as the most prevalent pavement distress, and the performance of these models has been evaluated considering learning and test speed, and accuracy as the most important performance parameters. The results of this research indicate that the speed of models almost depends on the pre-trained models characteristics that applied in transfer learning process. Also the accuracy of models based on various metrics (Sensitivity, F-score, etc.) is in range of 0.94 to 0.99 and indicates that deep learning method can be used to create expert systems for detection, classification, and quantification of pavement distresses such as cracking.

The ability of recording the high resolution spectral signature of earth surface would be the most important feature of hyperspectral sensors. On the other hand, classification of hyperspectral imagery is known as one of the methods to extracting information from these remote sensing data sources. Despite the high potential of hyperspectral images in the information content point of view, there are a numerous challenges in reliable extraction of information from these images. The issues such as 1- spectral similarity of different phenomena, 2- sensor noises and atmospheric effects, 3- the effects of high dimensionality in the pattern recognition algorithms, 4- the necessity of large number of training data to perform a reliable classification, and 5- spectral variability of similar phenomena could be considered as some of the challenges in hyperspectral data processing. Decreasing of the high dimensionality effects via the dimension reduction algorithms (e.g. band selection and feature extraction algorithms), as well as increasing the separability of the overlapped classes through the linear/non-linear mappings into the feature spaces with the higher dimensional are two opposite and conventional approaches of hyperspectral data processing. These approaches would be used based on the factors such as 1- complexities of classes in the imaging area, 2- spectral range of imaging sensor, and 3- the restrictions of processing algorithms. In this paper the fusion of these two approaches is used to perform an accurate hyperspectral image classification. To do so, a novel feature extraction method is proposed to be used in the hyperspectral image classification. The core of this method is the fusion of the linear, non-linear and sparse representation based features which is used to produce the effective features in the weighted K-Nearest Neighbors (KNN) classification method. In this procedure, a set of supervised and nonlinear features are extracted as the first step through the Nonlinear Principal Component Analysis (NLPCA). The supervised usage of NLPCA in order to extract features is known as one of the novelties of this paper. In this step, the spectral bands are usually mapped to a high dimensional feature space through the self-estimator artificial neural networks (ANNs) which are trained separately by ground truth data. In the second step, the previously extracted features are linearly transformed by the Linear Discriminate Analysis (LDA) method in order to reduce the dimension of the hypercube generated via supervised NLPCA to a separable feature space. In the last step, a set of features which is proportional to the number of classes is generated based on the sparse representation theory. The sparse representation features were hired to handle the effects of the inter-class variability. The precisions of the classified features in the two different hyperspectral images were on average shown 6 percent improvements in comparison with the spectral bands and the other combinations of extracted features. Furthermore, reach to the approximately 99% overall accuracies in the classes with the few training data could be considered as other achievements of the proposed method.

Land cover is defined as the physical material of the surface of the earth, including different vegetation covers, bare soil, water surface, various urban areas, etc. Land cover and its changes are very important and influential on the Earth and life of living organisms, especially human beings. Land cover change monitoring is important for protecting the ecosystem, forests, farmland, open spaces, water quality, wildlife habitat, and human health. Uncontrolled changes in land cover (deforestation, extensive agricultural activities, vast urban and exurban development, etc.) will not only destroy the natural environment and the ecosystem but also will affect different aspects of human life including the economy. As a result, various effective methods for land cover monitoring are always deliberated by researchers. Remote sensing based techniques, due to their unique capabilities and extensive data availability, have attracted great interest for decades. In addition, fully Polarimetric Synthetic Aperture Radar (PolSAR) images, due to the rich data and the ability of day-night and all weather condition data acquisition, has got high capability to study and monitor land cover changes. Recent developments of SAR sensors in acquiring fully polarimetric, continuous, and very high resolution data, besides enabling precise land cover monitoring, has necessitates the usage of more powerful and robust algorithms in different image processing steps. Classification algorithms are among the most fundamental tools of remote sensing image processing; and an increasing number of researches have been focused on developing novel and more robust classification methodologies for remote sensing data in recent years. Mid-level representations have also been introduced to the remote sensing data classification to enhance the reasonability and semantic rationality of the land cover classifications through remote sensing techniques. Bag of Visual Words (BOVW) model, originally inspired by the Bag of Words (BOW) model from text mining, is one of the most authoritative mid-level representation models which has achieved state-of-the-art results in different image processing tasks. BOVW representation model has recently been introduced to remote sensing data processing and proved to be very beneficial for high resolution remote sensing data representation. In the present paper, a novel segment-based BOVW framework has been developed for PolSAR data representation and classification. The obtained classified maps are quantitatively and qualitatively compared with the result of classification with well-known classifiers including Support Vector Machine (SVM), Artificial Neural Network (ANN), and Wishart Classification Algorithms. The main contributions of the current study are (i) covering the semantic gap between low-level features extracted from the PolSAR image and high-level concept of land cover in remote sensing data classification (ii) introducing novel segment-based BOVW framework for PolSAR image classification, and (iii) extensive evaluation of the performance of the BOVW model for PolSAR data classification in terms of the model parameters and low-level features selection. The experimental dataset is acquired by RADARSAR-2 from San Francisco Bay, California, in C band and at fine quad-pol mode, in 2008. The achieved overall accuracy, 90.1%, illustrates the capability of the proposed model for SAR image classification purposes. Moreover, the proposed segment-based framework for the BOVW model has decreased the undesired speckle effect of the SAR data, without speckle filtering.

The spatial analysis of accidents occurring in Highway with the aim of identifying theeffective parameters on the increase in the severity of accidents can be effective in decision making by professionals and decision makers to improve road safety in order to reduce the severity of road accidents.The purpose of this study is to provide a method for analyzing the severity of crashes and determining its effective factors in the freeways based on the spatial clustering functions and the data mining model of the regression tree.The proposedmethodis evaluated and tested on the Qazvin-Loshan freeway.In order to investigate the spatialdistributionofcrashes inthe axis of the study during the period of from2011to2016,the Geographic Information System’s spatial functions such as Getis-OrdG*Autocorrelation andkernel density functions have been Used.The preliminary results of spatial analysis showed that the focus of accidents in horizontal curves was greater.According to this achievement, in the next phase of the study, in order to study the factors affecting the severity of accidents, the Classification andRegressionTree was used on accidents occurringinthewholeaxisandspecifically in thehorizontal curveswith highcrashes.Thestudy of the relative importance of other variables in the proposed model shows that the geometric design,type of crashes and day of theaccident are the effective factors in increasing the accident severity at the Qazvin-Loshan Freeway.Inaddition,the results of modeling on horizontal curves crashes showed that the accident location,especiallytheshoulder,as well as the time of the accident,cause serious accidents.The results showedthattheintegrationofGISspatialfunctions withnon-parametricdecision-making tree-based datamining analysis,which is capable ofsimultaneous modelingofquantitativeandqualitativedata,is used to determine the factors affecting theseverityofaccidents andtospatial analyzethe patternsofaccidents usedinsuburbanarea,iseffective.

Improvement of the pavement surface texture charactristics and drainage quality is an important issue in the field of increasing roads safety and reducing the rate of accidents, especially in rainy weather conditions. The assessment of the pavement surface features and its relation with the accident rate is a common research topic, but no extensive research has been carried out on the evaluation of the pavement surface drainage. In this research, a system has been developed to assess the surface drainage of the pavements. For this purpose, a hardware is designed which is able to saturate the surface and capture the drainage process under constant conditions without the effects of environmental factors. The basis of the presented system is based on digital image processing techniques. Using image processing methods, three time-related indexes including Entropy,Energy and pixels proportion have been determined for the assessment of surface drainage quality of the pavement. Providing a proper combination of the indexes, the pavements are classified as appropriate, acceptable and inappropriate in terms of surface drainage performance using C5.0 datamining algorithm. The validation of the results of the proposed system shows that this system can evaluate the surface drainage situation with 95.7% accuracy. The presented system results can be used in the pavement management systems at project and network levels as a suitable measure for the evaluation of pavement safety in the rainy conditions as well as improving roads safety.

Form of the source-destination matrix Conventional approaches to accessing this matrix An essential input to the transportation planning process is the amount of travel demand in the directly, based on interviews and surveys in specific locations, are costly and time consuming. Hence, for the sake of mathematical methods based on intuitive statistics such as the volume of traffic in the network passages are becoming more and more popular in transportation studies. These methods are defined by having a source-primary-destination matrix as well as the current traffic flow information observed in a number of network paths, resulting in matrix estimation with the least distance from the primary matrix. , When allocated to the network, reproduce the observed volumes. Because conventional methods of obtaining this matrix that are based on interviewing and surveying are expensive, time consuming and annoying to people, the mathematical methods based on the observed statistics (such as the volume of network passage counts) ) Are expanding day by day. In this research, the source-destination matrix estimation methods can be divided into crowded and backbone networks in terms of network performance and the characteristics of each are presented.

Hyperspectral sensors have high capability in identifying objects by acquiring a large number of adjacent electromagnetic bands. Although This large number of bands makes it possible to approximate the more precise spectral curve of the material, it also brings some challenges. The difficulty in data transfer, the weak performance of conventional statistical classifications due to the limited number of training data, and the high processing time are the most important ones. Hence, different methods of dimensionality reduction are proposed for hyperspectral images. In the following article, an unsupervised feature extraction method is proposed based on the bands clustering technique. In the proposed method, after the prior image clustering and forming the prototype space with the aid of the clusters’ averages, the bands are clustered using the K-medoids clustering algorithm. In each cluster, four types of central tendency measures, mean, geometric mean, harmonic mean, and median are used to extract the final features. The experiments are conducted on the three real hyperspectral images with medium and high spatial resolution. Final results indicate that the classification results of the proposed method can reach (72.12) which is 7% higher than the other four competitive methods, principal component analysis (PCA) (64.39), wavelet (64.58), feature selection method based on bands clustering based on variance (65.30) and non-parametric weighted features extraction (NWFE) (64.12) .

Usually indicators like number of crashes and possible costs related to it are considered for safety assessment while those indicators are not controller and have no effects on safety and sometimes this information are not accessible for understudied area. Therefore, other safety indicators are required for safety control and supervisory which are called surrogate indicators. Vehicle speed is a key factor in road traffic safety. Theme of current paper is identifying effective factors on speed violation as surrogate indicator in intercity roads of Isfahan province by using classification and regression tree (CART) model. Information of vehicle speed and violating road speed limit in 10 intercity roads axis of Isfahan province are gathered by using Robot speed cameras. Also 4 main category of factors affecting speed violation including human, road, vehicle and environmental factors of those roads axis are gathered. Speed violation model has 74.7% of total accuracy. Results indicated that among 4 main categories of factors affecting speed violation, human factors have the most impact on speed violations. Among these human factors, frequency and duration of rests, driver experience level, travel time, driver literacy level, talking on phone while driving and presence of second driver in vehicle have more importance in the model classification improvement. After human factors, road factors are more important than other factors in classification of speed violation in which number of attractive points and volume have the most impact. In contrast to above mentioned factors, vehicle related a factor shows the least importance to model classification.

Information of the land uses is important for many activities, including the planning and the land management such territorial planning changes from recent decades to now. In this research has been afforded that the different land uses changes to be investigated and analyzed. For this purpose, using the satellite images from 1991 to 2016 and using the classification method of maximum likelihood, the map of land uses for the mentioned years was prepared. Then with LCM program, the changes of the different land uses from 1991 to 2016 to be analyzed. Results showed that the uses of rainfed lands by a minimum of 352 ha and a maximum of 365 ha and 717 ha total changes had the most changes, and the residential land by a minimum of 1 ha and a maximum of 56 ha increased and 57 ha total changes had the fewest changes among the uses studied. Furthermore, the general monitoring of changes showed the decrease of 506 ha pastures and increased 344 ha of the irrigated agriculture. Results showed that the natural areas was destructed and transformed into human-made areas. It is suggested that in order to conserve the natural resources, take decisions by the managers in planning

The last few decades witnessed high urban growth rates in many countries. Urban growth can be mapped and measured by using remote sensing data and techniques along with several statistical measures. The purpose of this research is to detect the urban change that is used for urban planning. Change detection using remote sensing images can be classified into three algebra-based, transformation-based and classification-based. By using any of these methods and applying them to SAR and optical data has advantages and disadvantages. Fusion of these methods and datasets can give us this opportunity to overcome their disadvantages and complement each other. For this purpose, here, a decision level fusion technique based on the majority voting algorithm is proposed for integrating the change maps extracted by different methods. After extracting features for optical and polarimetric data, object-based and pixel-based classification methods applied to optic images and also Wishart and SVM classification methods applied on SAR data. Change maps extracted from applying different CD methods such as post-classification, image differencing and principal component analysis. In order to evaluate the efficiency of the proposed method, various optical and radar remote sensing images from before and after of urban growth, acquired by QuickBird and UAVSAR, were utilized. In order to clarify the importance of using both optical and polarimetric images, the majority voting fusion algorithm on the change maps extracted by optical and polarimetric images was also applied separately. The results show that by fusing optical and polarimetric data at the decision level, it is possible to obtain a better accuracy because these two types of data, due to differences, can detect changes in a different way, thus covering each other's deficiencies. Polarimetric images better detect changes in altitude changes, and optical images better detect changes resulting from spectral changes. In order to perform a comparative evaluation, the accuracy of the change map obtained using optical images (total accuracy: 80.86% and kappa: 0.67), polarimetric images (overall accuracy: 75.43% and kappa: 0.5), simultaneous applying both datasets (overall accuracy: 88.48% and kappa: 0.79), as well as using the change maps of both data sets with the highest accuracy (overall accuracy: 88.81% and kappa: 0.79) have been obtained. In the end, due to the noise characterization of the post-classification method, the obtained change map improves with an overall accuracy of 90.11% and a kappa of 0.82.

Hyperspectral images are widely used in various fields such as agriculture, geology and mining, urban management, military, target detection. Classification is one of the most important fields of hyperspectral data processing which traditionally performed with spectral features. Various studies have shown that the use of spatial features along with spectral features can increase the accuracy of the classification. In this research, a method has been developed for the spatial-spatial classification of hyperspectral images. In this method, after a feature extraction step based on the minimum noise fraction (MNF) method, the geometric moments as spatial features are generated from first few MNF components in the different window seizes. In the next step, these features are stacked with spectral features and genetic algorithm is used to select the appropriate combination. Finally, a majority voting filter is used in order to remove the wrong pixels in each class, and increasing the homogeneity of labels and classification accuracy. Using the MNF transform to generate geometric moment features of the image, the use of genetic algorithm for selecting appropriate spectral-spatial features and post-processing step based on the majority voting filter are the innovative points of this paper. The results of the implementations on a real hyperspectral image from the semi-urban agricultural area named Indian Pines show that the proposed algorithm can reach the classification accuracy above 94% which is 40% better than conventional method.

From 1960s several attempts have been made to measuring the rock brittleness index BI. Schwartz (1964) using results of a series of triaxial tests on rock samples, stated that the rock’s behavior from frangibility to ductility happens in 4.3 ratios of principal stresses. Altindag (2002; 2003) introduced a new method for prediction of the BI by the division of the uniaxial compressive strength (UCS) of the rock to Brazilian tensile strength (BTS). In the late 1960s punch penetration test (PPT) introduced by Handewith (1971) to measure some physical properties of rock sample related to hardness and toughness of rock. Yagiz (2006) stated that the PPT’s results for measuring the BI have a very high correlation with TBM penetration rate. Although the PPT has very delightful results, application of this test is very expensive and needs much time as well.
Since 2002, researchers have made some efforts to predict the BI results acquired by PPT. Yagiz (2009) using 48 sets of 30 different tunnel rock’s PPT test data, introduced a new linear multivariable equation for prediction of the BI. In that equation, 3 major rock’s physical properties as UCS, BTS and unit weight were used. Also, Yagiz (2010) introduced a new nonlinear multivariable equation and improved the accuracy of prediction from R2=0.88 to R2=0.89. As well Khandewal et al (2016) developed a new equation using Genetic Programming (GP) based on Yagiz (2009) data with R2=0.90, but none of these equations include the type of rock as a major factor that influences the BI. In the present study, the BI will be investigated by considering the rock type (or texture) as an important parameter, and a new nonlinear multivariable equation will be introduced. As well using a classification and regression tree (CART) a new attempt will be made to predict the PPT’s result for the BI.