Please click on one of the flags to reset
ReadingDirection if you consider the current setting invalid
توفيق عبد الخالق عباس الاسدي
27/12/2012 06:42:05
تصفح هذه الورقة الالكترونية بتقنية Media To Flash Paper
Genetic Algorithm to find optimalGLCM features
Ruaa Mohammed Hamza Dr. Tawfiq A. AlAssadi
Department of Computer science College of Information Technology
Abstract
This paper presents a novel method to Image Retrieval Based on optimal Texture Features extracted from GLCM using Genetic Algorithm . The basic approach used here is that the textures features values that extracted from gray level cooccurrence matrix (GLCM) gives the typical values for features analysis .The Genetic Algorithm finds optimal Texture Features extracted from GLCM based on the fitness function . The obtained results of different types of images like "texture" "nontexture" and as unknown images where characterized in a good range.
Keyword: optimal GLCM features, GrayLevel Cooccurrence Matrix(GLCM), Texture features1.
Introduction
This paper introduces a new approach to find optimal GLCM features which are the most important in the field of image texture analysis using Genetic Algorithm (GA). The typical system performs three major tasks.The first one is texture analysis for features extraction, There are several number of texture analysis techniques that have been used in image processing area. Generally, the texture study includes: structural, transform method, and statistical model . The most common secondorder statistic is gray level cooccurrence matrix (GLCM) that used in this paper , the GLCM is essentially a twodimensional histogram of the number of times that pairs of intensity values (or , more generally , arbitrary local features) occur in given spatial relationship . thus , it forms a summary of the sub patterns that could be formed by intensity pairs and the frequency with which they occur .The second task is feature extraction (FE) from GLCM , For feature extraction in content based image retrieval there are mainly two approaches [5] feature extraction in spatial domain and feature extraction in transform domain [15]. The feature extraction in spatial domain includes theCBIR techniques based on histograms [5], BTC [4, 8, 9], VQ [3,10,11]. The transform domain methods are widely used in image compression, as they give high energy compaction in transformed image. So it is obvious to use images in transformed domain for feature extraction in CBIR [11, 12]. Transform domain results in energy compaction in few elements, so large number of the coefficients of transformed image can be neglected to reduce the size of feature vector.The third task is use Genetic Algorithm , where Genetic Algorithm (GA) is used for optimizing or selecting best features which gives a reduced feature set eventually results in high classification accuracy. Reducing the3dimensions of the feature space not only reduces the computational complexity, but also increases estimated performance of the classifiers. GA is biologically inspired and has many mechanisms resembling natural evolution [1] [6] [2].There are other approaches that used GLCM with Genetic Algorithm in computer vision , Jestin V.K. and J.Anitha used GLCM with GA for Retinal Image Analysis [14] , Lijun Qian and Jianrong Xu used for Improvement of Feature Selection in multiphase CT images of hepatic lesions [16]When we propose Genetic Algorithm to find optimal GLCM features, it is necessary to allocate following points.a. Divide image to number of blocks each with same size and give a label to each block .b. Each chromosome is used to represent a sort of block Matrix . This paper introduces a new approach to find optimal GLCM features which are the most important in the field of image texture analysis using Genetic Algorithm (GA). The typical system performs three major tasks.The first one is texture analysis for features extraction, There are several number of texture analysis techniques that have been used in image processing area. Generally, the texture study includes: structural, transform method, and statistical model . The most common secondorder statistic is gray level cooccurrence matrix (GLCM) that used in this paper , the GLCM is essentially a twodimensional histogram of the number of times that pairs of intensity values (or , more generally , arbitrary local features) occur in given spatial relationship . thus , it forms a summary of the sub patterns that could be formed by intensity pairs and the frequency with which they occur .The second task is feature extraction (FE) from GLCM , For feature extraction in content based image retrieval there are mainly two approaches [5] feature extraction in spatial domain and feature extraction in transform domain [15]. The feature extraction in spatial domain includes theCBIR techniques based on histograms [5], BTC [4, 8, 9], VQ [3,10,11]. The transform domain methods are widely used in image compression, as they give high energy compaction in transformed image. So it is obvious to use images in transformed domain for feature extraction in CBIR [11, 12]. Transform domain results in energy compaction in few elements, so large number of the coefficients of transformed image can be neglected to reduce the size of feature vector.The third task is use Genetic Algorithm , where Genetic Algorithm (GA) is used for optimizing or selecting best features which gives a reduced feature set eventually results in high classification accuracy. Reducing the3dimensions of the feature space not only reduces the computational complexity, but also increases estimated performance of the classifiers. GA is biologically inspired and has many mechanisms resembling natural evolution [1] [6] [2].There are other approaches that used GLCM with Genetic Algorithm in computer vision , Jestin V.K. and J.Anitha used GLCM with GA for Retinal Image Analysis [14] , Lijun Qian and Jianrong Xu used for Improvement of Feature Selection in multiphase CT images of hepatic lesions [16]When we propose Genetic Algorithm to find optimal GLCM features, it is necessary to allocate following points.a. Divide image to number of blocks each with same size and give a label to each block .b. Each chromosome is used to represent a sort of block Matrix . .
c. Texture Features extraction using GLCM Matrix . d. Calculate fitness function based on GLCM features .
Dear visitor, For downloading the full version of the research/article click on the pdf icon above.

وصف الــ Tags لهذا الموضوع
 Genetic Algorithm to find optimalGLCM features
