No abstract

Abstract:Feature selection is an important step in processing the images especially for applications such as content based image retrieval. In large multimedia databases, it may not be practical to search through the entire database in order to retrieve similar images from a query. Good data structures for similarity search and indexing are needed, and the existing data structures do not scale well for the high dimensional multimedia descriptors. Thus feature selection is an important step. Fuzzy rough feature selection method has many advantages in determining the relevant features. In this paper, five feature selection methods are compared with the fuzzy rough method. These five feature selection methods are Relief-F, Information Gain, Gain Ratio, OneR and the statistical measure 2 . The main purpose of the comparison is to rank the image features and see which method provides better results. An image retrieval dataset (COREL dataset) was used in the comparison. In order to evaluate the performance of the six methods, ranking of the important features is defined. This is then used to compare with the automated ranking produced by the aforesaid feature selection methods. Results show that the retrieval system using fuzzy rough feature selection has better retrieval accuracy and provide good Precision Recall performance. The advantages of the use of fuzzy rough feature selection will also be discussed in the paper.

Small wind turbines (SWTs) are often sited in more complex environments than in open terrain. These sites include locations near buildings, trees and other obstacles, and in such situations, the wind is normally highly three-dimensional, turbulent, unstable and at times weak. There is a need to understand the turbulent flow conditions for a small wind turbine in the built environment. This knowledge is crucial for input into the design process of a small wind turbine to accurately predict blade fatigue loads and lifetime, and to ensure that it operates safely with a performance that is optimized for the environment.

Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE. Feature Decreasing Methods Using Fuzzy Rough Set based on Mutual InformationMaryam Shahabi Lotfabadi , Mohd Fairuz Shiratuddin, Kok Wai Wong School of Information Technology, Murdoch University, 90 South Street, Murdoch, Western Australia.M.shahabi@murdoch.edu.au, f.shiratuddin@murdoch.edu.au, k.wong@murdoch.edu.auAbstract: Feature reduction methods are of interest in applications such as content based image and video retrieval. In large multimedia databases, it may not be practical to search through the entire database in order to retrieve the nearest neighbours of a query. Good data structures for similarity search and indexing are needed, and the existing data structures do not scale well for the high dimensional multimedia descriptors. Thus feature reduction is an important step. We investigate the use of rough set for feature reduction. In this paper, we compare three different decreasing methods. They are rough set, fuzzy rough set and fuzzy rough set based on mutual information. From the experimental results, it is shown that the fuzzy rough set based on mutual information can perform better than the other two rough set decreasing methods with increased image retrieval precision.Key words: Rough set, Content-based image retrieval, Fuzzy Rough set, mutual information. 1-IntroductionThe advancement in computing has produce innumerable digital images, photos and videos. This exponential growth has created a high demand for efficient tools for image searching, browsing and retrieval for use in various domains such as architecture, crime prevention, fashion, medicine, remote sensing, publishing, etc. This issue of large database has been addressed by an integrated framework called Content Based Image Retrieval (CBIR). Content-based image retrieval is one of the important topics in machine vision and research started as early as the 90s [1,2].In general, the dimensionality of image feature vectors used in image retrieval applications is quite high. Typical feature vector dimensions can range from few tens to several hundreds. For example, a colour histogram may contain 256 bins. This high dimensionality of the feature vectors creates problems in constructing efficient data structures for search and retrieval [3]. It is well known that most of the indexing structures do not scale well when the dimensionality of the feature vector exceeds 20. For this reason, there is considerable interest in reducing the dimensionality of the feature vectors [4].Previously investigated methods for feature reduction include the Principal Component Analysis (PCA) [5], Singular Value Decomposition (SVD) [15], Fastmap [5] and Multidimensional Scaling (MDS) [6]. In the last decade, researchers are working on the us...

In many researches, valuable studies have been done for feature extraction from images data-base, but because of weak classifiers using, good results have not been achieved. In this paper, different classifiers are compared in order to increase image retrieval system precision. Five different classifiers are used in the paper: the support vector-machine, the MLP neural network, the K-nearest neighbor, the rough neural network, and the rough fuzzy neural network. The rough fuzzy neural network and the rough neural network have not been used in image retrieval implication up to now. The innovation of this research is the using of these classifiers in the image retrieval implication. From the performed test, it is concluded that the rough fuzzy neural network classifier has performed better than other classifiers and increased the image retrieval precision. The COREL image data-base with 1000 images in ten content groups has been used and the classifiers have been compared.