clustering

Traffic monitoring and control in urban and intercity highways plays an effective role in increasing the efficiency of resources and infrastructure, as well as increasing the satisfaction of the stakeholders of transportation systems. Analyzing random traffic variables such as traffic and speed and diagnosing and predicting their unusual situations, which generally indicate the existence of traffic crises, is one of the approaches to monitoring traffic flow and solving traffic problems in metropolises. In this article, by using the data collected by traffic sensors, the data related to traffic variables have been grouped using an innovative algorithm based on the particle swarm optimization method in terms of the traffic situation. The grouped data were used to calculate suitable control limits for time intervals, and the results indicate the high performance of this method. For this purpose, in the first step, using an innovative method based on particle swarm optimization, the training data is grouped based on the values of traffic variables, including density, traffic and speed, and then the appropriate group is identified for each variable value. In the next step, the traffic situation is calculated for the clusters and hours of the day, and based on that, superior and standard control charts are drawn. The obtained results indicate the appropriate accuracy of the proposed system in traffic monitoring.

Estimating a structure's modal parameters is essential for various applications, including health monitoring, damage detection, design verification, and model updating. Modal parameters are a structure's natural frequencies, mode shapes, and damping ratios. They can be used to understand the structure's dynamic behavior and to identify any changes that may occur over time. Operational modal analysis (OMA) is a technique that uses the response of a structure to environmental loads to estimate modal parameters. OMA is a non-destructive testing method that can be used on structures in their operating environment. This makes it a valuable tool for health monitoring and damage detection of buildings, bridges, wind turbines, and stadiums. One of the challenges of OMA is that its methods rely on the user's judgment to separate physical modes from spurious modes and to distinguish between real modes of the structure. Spurious modes are not caused by the actual structure but by noise or other environmental factors. Real modes are caused by the structure itself. In recent years, there has been extensive research on automating OMA methods for modal parameter estimation. Most of these studies have attempted to minimize the need for user intervention in modal parameter calculation by using machine learning techniques. Machine learning techniques can be used to identify physical modes automatically and to distinguish between real modes of the structure. This research uses the Stochastic Subspace Identification (SSI) method for OMA. The DBSCAN clustering method is used to separate physical modes from spurious modes. Finally, the hierarchical clustering method is used to distinguish between real modes of the structure. The proposed algorithm was implemented on a 6-degree-of-freedom structure and a real bridge. The results show that the proposed method has a higher power to separate physical modes from spurious modes than previous methods.

Today, the interaction of companies with customers in the form of customer relationship management (CRM) has changed significantly. Identifying the characteristics of different customers and allocating optimal resources to them according to their value for companies has become one of the main concerns in the field of customer relationship management. The aim of this research is to provide a suitable model for customer segmentation based on their value creation. In the proposed process of this research, which has been implemented in a passenger rail transport company, the data related to all passengers and customers of the company in a period of 4 years; From the beginning of 2014 to the end of 2017, and after determining the values of the indicators of the model that is R. F. M. (Recency; Frequency, and Monetary value) customers are clustered based on three criteria, using K-means clustering technique. Next, by weighting the criteria (RFM) based on the best-worst method and calculating the lifetime value, the existing clusters are prioritized and the key and valuable customers of the company are identified. Finally, suggestions for improving the Customer Relationship Management system is presented.

Reducing the consumption of energy carriers is one of the main issues and necessities in the current economic conditions of the country. Natural gas is one of the important carriers that is responsible for most of the thermal energy supply in freight and passenger terminals. In this research, first, a set of solutions to reduce natural gas consumption in terminals have been collected, a significant number of which are related to the powerhouses of terminals. Then in this regard, five criteria of technical ease of implementation of the solution, ease of measuring the reduction of gas consumption through the implementation of the solution, appropriate time to implement the solution, access to specialized technicians to implement the solution and access to parts and tools needed to implement the solution. Evaluation and prioritization of solutions were considered. The weight and importance of each criterion are not equal. Therefore, pair-wise comparisons were made between the two criteria to finally calculate the weight coefficients of the criteria. These strategies were scored based on the above five criteria by a group of experts and were prioritized based on the scores obtained using the Analytic Hierarchy Process (AHP) technique. Finally, using the K-Means clustering technique, the solutions were classified and presented in three groups based on the scores of the five criteria evaluated. Some of these solutions are related to technical adjustments, insulation, equipment replacement, compliance with performance standards, installation of new equipment, and basic measures.

The occurrence of crime in society causes insecurity and chaos in people's lives, imposes extensive material and spiritual costs on society, and diverts a lot of time, money, and energy from the country's security agents. Spatial investigation of a crime in the past and predicting the likelihood of its occurrence in the future, or in other words, determining the spatial centers of crime, can help manage crime prevention better or deal with it effectively. Unfortunately, measuring a crisis center is a complex issue. In this regard, many statistical methods can be used to identify crisis hotspots, including cluster analysis. In the present study, for the proper efficiency of such analyzes, with a combined approach of spatial information systems and K-means clustering, the spatial distribution of crime in Qazvin (as a case study) is investigated. Assist in locating crime centers more accurately. The results showed that this combined approach could model the local centers of crime in Qazvin.

Excessive use of groundwater resources has put the aquifers in critical situations. This study provides a framework for using the Bayesian network for groundwater level estimation and aquifer hydrograph analysis. A 10-year statistical data, 8 years data for model training, and 2 years data for model validation, were used. The Bayesian network model was implemented and analyzed in three explicit, clustering, and two- and three-month delay states. Explicit simulation results showed that most of the wells have a good correlation between the simulation and observed data. Clustering results were less accurate than explicit ones. In the third case, two and three months delay data was used to simulations. The results showed that the correlation between observed and simulated groundwater levels decreased. At 1, 2 and 3 months delay statues, Root Mean Square Error was 1.87 m, 3.76 m, and 6.42 m, respectively. Therefore, the one-month lag time was chosen for the simulations and aquifer hydrograph was used to evaluate and estimate total aquifer variations. The results indicate the appropriate accuracy of the aquifer parameters estimation.

The efficiency of contractors selection in infrastructure projects is always one of the main concerns of employers. The purpose of this study is to propose a new method for specification of evaluation criteria, according to the type of project under assignment and prioritization, and to determine the proposed score of the evaluation criteria of the contractors using two different methods. The first involves statistical analysis of the questionnaires and the second involves the text mining of the interviews. For this purpose, after library studies, fieldwork and questionnaire design, questionnaires were given to experts in the industry and their statistical analysis identified important criteria and sub-criteria and then interviewed another group of experts. Using the text mining and clustering of interview texts, the criteria and indices of contractors' evaluation were identified and the results of these two methods were compared with the bidding law implementing regulations in Iran. SPSS software was used to analyze the interviews and K-means algorithm was used to interviews text mining. The findings of the study indicate that the results of the text mining and questionnaire are in agreement and It can be concluded that the identified criteria have acceptable accuracy and generalization capability and organizations can use these criteria and evaluation indicators to select the most suitable contractor in the tenders. Identified Criteria, in addition to covering all the criteria introduced in the Bidding Law Implementing Regulations, also include two new criteria, namely Claim Management and Safety Management & Quality Control.

In recent years, the advancement of information gathering technologies such as GPS and GSM networks have led to huge complex datasets such as time series and trajectories. As a result it is essential to use appropriate methods to analyze the produced large raw datasets. Extracting useful information from large data sets has always been one of the most important challenges in different sciences, and data mining techniques provide useful solutions to solve this problem. Nowadays, clustering technique as the most widely used function of data mining, has attracted the attention of many researchers in various sciences. Due to different applications, the problem of clustering time series data has become highly popular and many approaches have been presented in this field. An efficient clustering method groups data in such a way that the objects in the same cluster are more similar to each other than to objects in different clusters. In order to compute the difference/similarity between time series data in clustering process, a similarity measure or distance function is used. Therefore, choosing an appropriate distance function is one of the most important challenges that should be considered before starting the clustering process. So far, various distance functions have been proposed to measure the difference/similarity between time series and each of them have its own strengths and weaknesses. Since choosing a suitable distance function to cluster a specific data set is a complicated process, in this study, we proposed a clustering method based on combination of the well-known Fuzzy C-Means (FCM) method and the Particle Swarm Optimization with the ability of using different distance functions in time series clustering process. In this way, the step of choosing the best distance function before starting time series clustering procedure has been deleted and different similarity measures can participate in the clustering process with different impacts. The objective function in this study is defined based on Fuzzy C-Means clustering objective function and the particle Swarm Optimization algorithm is used to find the optimal value for the considered objective function. Finally, by considering three distance functions including Euclidean distance, dynamic time warping and Pearson correlation coefficients the proposed method was implemented on seven well-known UCR time series datasets. Also, by considering the average normalized mutual information as a criterion for evaluating the performance of methods in this research, the proposed method was compared with five other methods. The results of this comparison indicated that the method presented in this study performed better in more than 85% of cases rather than other methods. In order to have a better evaluation, Tukey’s multiple comparison tests with a threshold of p < 0.05 is used with the ability of comparing the methods in pairs. The results obtained by Tukey test showed that, in about 83% of cases, the difference between achieved results by the proposed method in this study and results obtained by the other five techniques are statistically significant. Overall, the results of this study clearly showed the superiority of the proposed clustering method in the production of high quality clusters in comparison to some other methods.

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) .

Cable-stayed bridges are vital structures which need significant maintenance and repair costs every year. Therefore, health monitoring of such structures can mitigate human and financial losses. In this paper, a damage detection method for cable-stayed bridges is proposed using signal processing and clustering. Since the accuracy of signal processing can considerably affect the accuracy of damage detection results, in the first part of the paper, a comparison is carried out between the popular FDD method and two newer AFDD and TDD methods which have improved some of the FDD drawbacks. Then, the most effective method is selected. Among these procedures, FDD is successfully implemented in signal-based procedures. However, the two newer ones have not adequately investigated in comparison to FDD. In the second part, by using competitive neural network for clustering, a new damage index is introduced by calculation of the Euclidian distances of cluster centers. Results show that the proposed damage detection algorithm can differentiate healthy and damage states with acceptable accuracy.

In recent years, the tremendous and increasing growth of spatial trajectory data and the necessity of processing and extraction of useful information and meaningful patterns have led to the fact that many researchers have been attracted to the field of spatio-temporal trajectory clustering. The process and analysis of these trajectories have resulted in the extraction of useful information which is able to respond to different challenges in real-world applications such as traffic management and control, smart transportation, surveillance, security and biological studies. Clustering is one of the most important methods for trajectory pattern extraction, their volume reduction, discovering outliers in trajectories, indexing and their simple visualization. So far, different similarity functions and clustering algorithms have been proposed for trajectory clustering. The diversity of clustering algorithms and their unique results highlights the need for paying attention to their weaknesses and strengths. Some clustering algorithms are only effective on low volume datasets. There are also some algorithms which are only able to extract clusters with convex shape, whereas some of them extract clusters of any shapes. On the other hand, several clustering functions require the determination of the initial value, such as the number of clusters by the users while some others do not need initial inputs. In addition, outlier detection is not possible in all clustering algorithms. In this study, spatial trajectories clustering algorithms that are extended from point clustering algorithms is divided into four general categories: partitioning-based clustering, hierarchical clustering, optimization-based clustering and density-based clustering. Then, the most commonly used algorithms in each category are implemented and evaluated. The evaluation process is performed on two sets of data (cross and i5) with dissimilar complexity. The effect of noise and outliers is one of the most critical parameters engaged in the performance quality of clustering functions which is considered in this study. The Silhouette index and computational time are used as two parameters for comparison and evaluation. According to obtained results, it is crucial to consider the data, its features, and also the utilized distance function in order to decide on the proper clustering method.  However, generally, the best results regarding the clustering quality are obtained from optimization-based clustering. With the integration of genetic algorithm into the K-means, all results in two cases of using both two datasets and using two different distance functions are improved. Using the genetic algorithm in K-means leads to finding the optimum location of cluster centers and dealing with the local minimum problem. It is important to note that high computational time is one of the weaknesses of optimization-based clustering. After the optimization-based clustering, regarding the clustering quality, partitioning-based, hierarchical and density-based clustering have achieved the second, third and fourth ranks respectively. With regard to the computational time, the best results are obtained from the density-based, hierarchical, partitioning-based and optimization-based clustering consecutively. Some methods such as K-means (a sub-category of partitioning-based clustering) are severely sensitive to outliers while spectral sub-category of partitioning-based clustering has a high resistance against them. Moreover, the density-based and optimization-based clustering methods have the highest tolerance against noise.

Iran is known as one of the high risk seismic regions of the world. Over the past 50 years, many destructive earthquakes have occurred in this area, causing much human loss and financial damage. So, from the perspective of emergency-management and hazard preparedness, it is essential to make an effort to predict earthquake occurrence. Earthquake prediction is an instance of interdisciplinary research, which is a concern of many scientists in various fields, such as geology, seismology, engineering, mathematics, computer science and even social sciences, who study different aspects of the matter to find new solutions. Efforts in this field are divided into long-term and short-term predictions. The short-term predictions are based on precursors such as foreshock, seismic quiescence, decrease in radon concentrations and other geochemical phenomenon. Due to numerous complexities and unknown factors inside the earth, exact prediction of earthquakes is difficult and practically impossible. During the last two decades, many techniques have been developed to discover the pattern of seismic data and predict three earthquake parameters, namely; time of occurrence, location and magnitude of future earthquakes. Soft computing and data mining techniques, such as neural networks, fuzzy logic and clustering methods are appropriate tools for problems, such as earthquake prediction, that suffer from inherent complexities. Many researchers have used these approaches to reduce uncertainty in results. In this paper, the b-value of the Gutenberg Richter law has been considered as a precursor to earthquake prediction. Prior to earthquakes equal to or greater than $M_w$ = 6.0, temporal variation of the b-value has been examined in Qeshm island and neighboring areas in the south of Iran, from 1995 to 2012. The clustering method, by the k-means algorithm, and a self-organizing map (SOM) have been undertaken to find a pattern of variation of the b-value. Three clusters are obtained as an optimum number of clusters by the Silhouette Index and the Davis-Bouldin index. Prior to all the mentioned earthquakes $(M_wgeq 6.0)$ a cluster, known as a decrease in b-value, has been seen; so, a decrease in the b-value before main shocks has been considered as a distinctive pattern. Also, an approximate time of decrease has been determined.

Time series is a type of data with complex structure. Analysis of time series is used in sciences such as meteorology, economics, geology, marine science, medicine and engineering widely. So, Because of time series applications in various sciences, the interest to analyze these data has been increased.On the other hand by developing information gathering technologies such as mobile, GPS and sensors, and Access to large volumes of time series data, we always require methods to extract useful information from large datasets. Thus, data mining is an important method for solving this problem. Clustering analysis as the most commonly used function of data mining, has attracted many researchers in computer science. Clustering is a strong instrument for knowledge discovery and it provides useful information about existing patterns in datasets. In general, the purpose of clustering is representing large datasets by a fewer number of cluster centers. It simplifies large datasets and thus is an important step in the process of knowledge discovery and data mining. Fuzzy C-means (FCM) clustering is one of the most important classic clustering methods that have been used in many researches. The main disadvantage of this method is the high probability of getting trapped in local optima especially in facing high-dimensional data such as time series. Furthermore Euclidean distance is the most commonly used similarity measure in Fuzzy C-means but sometimes, its necessary to use another similarity/dissimilarity measures instead of Euclidean distance. In this paper in order to compensate the shortcomings of Fuzzy C-means algorithm, we used one of the existing evolutionary algorithms. Evolutionary algorithms has gained huge popularity in the field of pattern recognition and clustering recently. Among the existing evolutionary algorithms, the differential evolution algorithm as a strong, fast and efficient global search method has been attracted the attention of researchers. In this paper, we proposed a technique for clustering time series data using a combination of Fuzzy C-means and differential evolution (DE) approach and we considered dynamic time warping (DTW) as distance measures between time series. Also, in this method we used Discrete Cosine Transform (DCT) to time series dimension reduction. Finding all elements of cluster centers using differential evolution is time consuming and the large number of unknown parameters related to the cluster centers will reduce the efficiency and the speed of differential evolution algorithm.So, for reducing the search space,the most important Discrete Cosine Transform coefficients of the cluster centers were recognized as the main unknown clustering problem in the proposed method and differential evolution algorithm tries to determine the near optimal Discrete Cosine Transform coefficients of cluster centers by minimizing the Fuzzy C-means objective function. Experimental results over two popular data sets indicate the superiority of the proposed technique compared to fuzzy C-means and a clustering algorithm based on differential evolution without using dimension reduction techniques.Comparing the run time of the methods, the proposed method is slower than the Fuzzy C-means clustering algorithm, but due to the use of discrete cosine transform method to reduce unknowns, it operates faster than differential evolution without using dimension reduction techniques.

The occurrence of earthquake has made human being consider fundamental plans to reduce the consequent danger and destruction. The only means to reduce the vulnerability is to set specific management of urban crisis in construction; moreover, this aim cannot be achieved unless the city immunity, in confrontation of the earthquake, is considered as a major purpose in all stages of urban planning. Proper allocation of various urban land uses helps hugely the process of management of urban crisis related to the earthquake; accordingly, recognizing the different effective variables of vulnerability of urban areas from the aspects of urban land use, definition and declaring their relations with vulnerability, their analysis and finally preparing the land use optimizing maps with less percentage of vulnerability, is the principle target of this paper.
In this paper, for optimizing urban land use allocation, with the approach of reducing the vulnerability caused by the earthquake based on the physical factors, the multi-objective optimization algorithm NSGA-II was used for modeling. The 12th district of Tehran was taken as the subject of study. In this algorithm the main objectives include: maximizing compatibility of adjacent land uses, accessibility of land uses, availability of sanitary-medical and residential land uses to the Road network and minimizing susceptibility in earthquake's time and Minimizing land uses change. Considering the fact that the NSGA-II algorithm is multi-objective, the decision maker encounters different solutions in the Pareto-optimal front, which makes the process more complicated. Accordingly, to aid the decision making process and presenting the correspondent scenarios with the decision maker's priority, the clustering analysis was used with K-means approach. To study the changes of the results of different implementations of algorithm and stability of optimization algorithm, convergence trend and repeatability test carried out.
In the resulted optimized land use arrangements, the levels of objective functions are much better than the previous arrangement. Moreover, accessibility objective function has been improved mostly under the effect of optimization (27 %). The average percentage of the improvement of the objective functions in the algorithm was 19 %. In the repeatability test, the average percentage of the overlay of the algorithm's solutions in different runs was recorded as 76 %, which can be recognized as a proper value, and represents the suitable repeatability of the algorithm. The results were found acceptable based on the convergence trend, by having the stable value of the objective functions after specific times of iteration.
Several factors represent the efficiency of the model which can be named as; the proper method of optimization that was compatible with the problem, defining the objective functions based on the reality and including the main aspects of the problem of the earthquake vulnerability in the presented model, concerning the opinion of the decision makers in the process of the research and the final stage for selecting the optimum arrangement with the analysis of the results of the scenarios and the scenario's clustering. The results of this research can be an aid as a means to support deciding for the planner and urban management policy makers encountering earthquake, in planning appropriately for the urban spaces.

LiDAR has been recently known as a powerful technology for 3D information acquisition from objects, and researchers are working on developing reliable techniques such as clustering for automatic object extraction from LiDAR data. Finding optimal clustering method, number of clusters and descriptors are the most challenging problems in clustering of LiDAR data. In this research, a GA based solution is presented for finding optimal descriptors of in LiDAR data clustering.
A vast number of features were considered in two categories of single operator and composite descriptors. Single operator features are extracted from first and last pulses of intensity and range data (FR, LR, FI and LI) in which, each operator are executed on this four data to extract four operator features. Besides, two other features of NDIR and NDII are also extracted from the resulted features. Here, ten groups of single operator features were extracted: Raw data (6 features), Scene Ratio(12 features), fractal (12 features), Gabor (12 features), Haralick (24 features), Moments (24 features), SemiVariogram (12 features), Normal Vector (36 features) and Roughness (6 features). Composite features are extracted using a combination of the operators in the first category which nine features were generated. Finally, 171 features are extracted from LiDAR data.
To simplify the computations, in the first level, unsuitable features are deleted from the list based on manual observation. Minimizing the NCE measure in clustering of LiDAR data using k-means algorithm, optimal descriptors in the remained list are assessed through Genetic Algorithm.
To evaluate the proposed method a sample area from the city of Stuttgart, Germany, were used. From the 171 extracted features 100 unsuitable features were deleted manually and the optimization process were executed on 71 features. In the sample area three clusters of Buildings, Trees and Land were considered for clustering. After executing GA process 5 features were selected as optimal features. Obtained results proved the ability of the GA algorithm as an optimization tool as well as the optimality of the selected descriptors in clustering.

In urban space, the need to different facilities and diverse land uses increases continuously. Continuous changes in the demands of the citizens results in rapid and frequent land use changes. Therefore, dynamic characteristics of urban environment should be considered in urban planning. On the other hand, land uses have different effects on each other. In other words, any change in the land use of a parcel or zone, will results in tendency of its neighboring parcels for land use change. Therefore, proposing of new land use arrangements after any occurred land use change could be a proper response to such tendencies. The main goal of this study is to propose a method, based on GIS and NSGAII optimization algorithm, for generating optimum land use arrangements after any occurred land use change.Usually, the output of a multi-objective optimization algorithm is a collection of optimum solutions. Selection of appropriate solution from such a collection needs extra efforts and processes. Therefore, another goal of this research is to use an appropriate clustering method that helps the user to select the most preferred solution. With such a method, the decision maker can introduce his planning priorities, perceive the resulted scenario and select accordingly. In this research, ant colony clustering algorithm is used for clustering, first because of its high speed, and then because the representative of the clusters are selected from the Pareto front solutions. In this research, for modelling of the aspects of land use change and its factors, four objective functions are considered, which are: the maximization of land use compatibility, the maximization of land use dependency, the maximization of land use suitability, and the maximization of land use compactness. Finally, the providence of per capita for different land uses is considered as constraints. The ant colony clustering algorithm is used for clustering of the found solutions (land use arrangements). The developed method is implemented and tested using the data related to some districts of region 7 of Tehran.Different evaluations are considered and carried out for the results of optimization. These include the convergence trend, repeatability test, and the comparison of the previous land use arrangement with the optimized ones. In the resulted optimized land use arrangements, the levels of objective functions are much better than the previous arrangement. Furthermore, the required per capita for different land uses are much better satisfied. The highest improvement in the objective functions is 36%, which is related to land suitability. In addition, the required per capita is improved by 18.5% of. The results of clustering using ant colony clustering algorithm are compared with those of K-means and Fuzzy K-means. The comparison showed that the ant colony clustering algorithm is faster. In addition, the results of this clustering method are exactly the original solutions of the land use arrangement optimization.Finally, the developed method can help urban planners and decision makers to correct and change the detailed urban plans according to any occurred land use change. One of the limitations of the detailed urban land uses plans is that they are not flexible and cannot opt to the deviations from the plan. This research is one step in the development of a general approach to dynamic urban planning. Such a planning approach can respond to the continuous and dynamic changes of the land uses in urban space.

Speckle phenomenon is appurtenance of radar image nature that is caused accuracy reduction in interpretation، classification، segmentation، etc. over these images. Researchers have tried to decrease this phenomenon in radar images. One of the speckle phenomenon reduction methods in the radar images is spatial domain filters. Main methods of the spatial domain filter use thresholds for classifying type of targets، then speckle phenomenon decreases based on type of target. Using threshold in images for target classification is a method with great error. Therefore، clustering method is used for target classification in this research. In proposed method، firstly، speckle phenomenon of radar raw image decreased by using Mean filter. Afterwards، K-Means clustering method implemented over the filtered image with various cluster numbers. Optimal cluster number in case study determined using Davis Bouldian index. Finally، filtered image was produced in a decision level with the clustering and the raw radar images. Results of the proposed method were compared with other method including Mean filter، Lee filter، Enhanced Lee filter، Gamma filter and Median filter by using radiometric and edge preservation indexes. Results show that the proposed method has higher accuracy than the other implemented methods based on the mentioned indexes.

In order to enhance the accuracy of the image clustering, many clustering algorithms such as semi-supervised clustering, clustering ensembles and Coclustering have so far been proposed where the most effective strategy is when spatial context in clustering is considered. In this study, four clustering methods in which the spatial neighborhoodinformation is used are investigated. Theseinclude: using texture information, object based clustering, Markov random field clustering and applying majority filter onthe clustering results. These four methodsare tested on the two synthetic and two real high resolution satellite images. The results show that using texture informationin clustering can enhance the overall accuracy of the clustering of the real images up to 10.8% relative to the simplek-means clustering. Markov random field clustering of the synthetic images shows 22.8% improvement in overall accuracy.The results show that using spatial context can enhance the overall accuracy of the satellite image clustering.