List of Articles خوشه‏‌‏بندي

• Open Access Article
  • Abstract Page
  • Full-Text
  
  1 - Outdoor Color Scene Segmentation towards Object Detection using Dual-Resolution Histograms
  javad rasti monadjemi monadjemi abbas vafaei
  One of the most important problems in automatic outdoor scene analysis is the approach of segmentation towards object detection. The special characteristics of such images -like color variety, different luminance effects and color shades, abundant texture details, and d More

  One of the most important problems in automatic outdoor scene analysis is the approach of segmentation towards object detection. The special characteristics of such images -like color variety, different luminance effects and color shades, abundant texture details, and diversity of objects- lead to major challenges in the segmentation process. In the previous research, we proposed a k-means clustering algorithm in a multi-resolution platform for preliminary color segmentation. In this method, the texture details are deliberately expunged and apparent clusters are gradually removed in the blurred versions of the image to let more detailed classes expose in the more clarified versions. The performance of this step-by-step approach is relatively higher than the traditional k-means in color clustering for outdoor scene segmentation. In this paper, an adaptive method based on the circular hue histogram in a dual-resolution platform is suggested to detect the apparent clusters in the blurred images. Experimental results on two outdoor datasets show about 20% decrease in the pixel segmentation error as well as around 30% increase in both precision and speed in the convergence of the clustering algorithm. Manuscript profile
• Open Access Article
  • Abstract Page
  • Full-Text
  
  2 - Segmentation of exterior color images for the purpose of object recognition using histogram with double accuracy
  javad rasti amirhasan Monjimie abbas vafaei
  One of the important issues in the automatic processing of external images is how to divide these images for the purpose of recognizing something in them. The special characteristics of these images, including color diversity, different light effects, the presence of co More

  One of the important issues in the automatic processing of external images is how to divide these images for the purpose of recognizing something in them. The special characteristics of these images, including color diversity, different light effects, the presence of colored shadows, many texture details, and the existence of small and heterogeneous objects, make the problem of segmentation of external images, especially color segmentation, face serious challenges. In previous researches, a method based on the k-means clustering algorithm was proposed in a multi-accuracy bed for color clustering of external images for the purpose of primary segmentation. This method uses deliberate blurring of image textural details and removal of specific classes in blurred images and then added The classification of classes in images with higher accuracy showed a suitable performance for the initial segmentation of these images in comparison with the normal k-means method. In this article, an image-adaptive method using the ring histogram of the dark color to identify specific classes in blurred images in the bed is presented. It has been proposed with double precision. The efficiency of this algorithm has been investigated with the help of a supervised evaluation method on two databases of external images, which shows a 20% reduction in pixel error in segmentation, as well as a 30% higher accuracy and speed in the convergence of the clustering algorithm, indicating a higher quality. The proposed method is better than the normal method. Manuscript profile
• Open Access Article
  • Abstract Page
  • Full-Text
  
  3 - Robustness of fuzzy c-mean method for delineation of hydrochemical facies distribution of groundwater in Varamin Plain
  Mohammad Nakhaei Mehdi Talkhabi Meysam Vadiati
  In this paper, classification of a large hydrochemical data set from Varamin plain is done by using fuzzy c-means (FCM) and hierarchical cluster analysis (HCA) clustering techniques. Then its application to hydrochemical facies delineation is discussed. Groundwater samp More

  In this paper, classification of a large hydrochemical data set from Varamin plain is done by using fuzzy c-means (FCM) and hierarchical cluster analysis (HCA) clustering techniques. Then its application to hydrochemical facies delineation is discussed. Groundwater samples were grouped into three classes according to the optimum number of the classes and fuzziness exponent by using the fuzzy c-mean. The data set includes 90 deep and moderate deep well samples from groundwater data set and 9 hydrochemical variables were used. Results from both FCM and HCA clustering produced cluster centers that can be used to identify the physical and chemical processes creating the variations in the water chemistries. The optimum cluster in FCM method determined by optimization function, but in HCA method by trial and error. The FCM method is potentially useful in establishing hydrochemical facies distribution and may provide a better tool than HCA for clustering large data sets when overlapping or continuous clusters exist. Plotting the cluster membership value contours on a map demonstrated the existence of three spatially continuous, well-defined clusters of groundwater samples. The results showed that the FCM method is more sound for investigating threshold data rather than HCA method (that represents sharp and abrupt variations). Manuscript profile
• Open Access Article
  • Abstract Page
  • Full-Text
  
  4 - Developing a hybrid model to clustering Tehran Stock Exchange companies using meta-heuristic algorithms
    Ali Mohaghar
  Investment decision, have always has been one of the most important issues. Investors are trying to achieve the highest efficiency and the least risk by selecting the best companies from Among a wide variety of companies considering to various financial indicators. Acco More

  Investment decision, have always has been one of the most important issues. Investors are trying to achieve the highest efficiency and the least risk by selecting the best companies from Among a wide variety of companies considering to various financial indicators. Accordingly, today, there are many ways to analyze the data from this company. One of the ways is clustering that classification of the companies. However, the present study aimed to identify and distinguish successful from unsuccessful companies in Tehran Stock Exchange has been done using K-means clustering. Then this problem is solved using meta-heuristic algorithms. The results indicate that meta-heuristic algorithms compared with conventional methods, more efficient and have led to a global optimum. Also these results of Altman’s bankruptcy model were confirmed results of meta-heuristic algorithms. Manuscript profile
• Open Access Article
  • Abstract Page
  • Full-Text
  
  5 - Improving Precision of Recommender Systems using Time-, Location- and Context-aware Trust Estimation Based on Clustering and Beta Distribution
  Samaneh Sheibani Hassan Shakeri Reza Sheybani
  Calculation and applying trust among users has become popular in designing recommender systems in recent years. However, most of the trust-based recommender systems use only one factor for estimating the value of trust. In this paper, a multi-factor approach for estimat More

  Calculation and applying trust among users has become popular in designing recommender systems in recent years. However, most of the trust-based recommender systems use only one factor for estimating the value of trust. In this paper, a multi-factor approach for estimating trust among users of recommender systems is introduced. In the proposed scheme, first, users of the system are clustered based on their similarities in demographics information and history of ratings. To predict the rating of the active user into a specific item, the value of trust between him and the other users in his cluster is calculated considering the factors i.e. time, location, and context of their rating. To this end, we propose an algorithm based on beta distribution. A novel tree-based measure for computing the semantic similarity between the contexts is utilized. Finally, the rating of the active user is predicted using weighted averaging where trust values are considered as weights. The proposed scheme was performed on three datasets, and the obtained results indicated that it outperforms existing methods in terms of accuracy and other efficiency metrics. Manuscript profile