Results 11  20
of
21
Color Image Segmentation with CLPSObased Fuzzy
, 2007
"... A new method for color image segmentation using fuzzy logic is proposed in this paper. Our aim here is to automatically produce a fuzzy system for color classification and image segmentation with least number of rules and minimum error rate. Particle swarm optimization is a sub class of evolutionary ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
A new method for color image segmentation using fuzzy logic is proposed in this paper. Our aim here is to automatically produce a fuzzy system for color classification and image segmentation with least number of rules and minimum error rate. Particle swarm optimization is a sub class of evolutionary algorithms that has been inspired from social behavior of fishes, bees, birds, etc, that live together in colonies. We use comprehensive learning particle swarm optimization (CLPSO) technique to find optimal fuzzy rules and membership functions because it discourages premature convergence. Here each particle of the swarm codes a set of fuzzy rules. During evolution, a population member tries to maximize a fitness criterion which is here high classification rate and small number of rules. Finally, particle with the highest fitness value is selected as the best set of fuzzy rules for image segmentation. Our results, using this method for soccer field image segmentation in Robocop contests shows 89% performance. Less computational load is needed when using this method compared with other methods like ANFIS, because it generates a smaller number of fuzzy rules. Large train dataset and its variety, makes the proposed method invariant to illumination noise. Key words: Comprehensive learning particle swarm optimization, Fuzzy
InformationTheoretic Image Reconstruction and Segmentation from Noisy Projections
"... Abstract. The minimum message length (MML) principle for inductive inference has been successfully applied to image segmentation where the images are modelled by Markov random fields (MRF). We have extended this work to be capable of simultaneously reconstructing and segmenting images that have been ..."
Abstract
 Add to MetaCart
(Show Context)
Abstract. The minimum message length (MML) principle for inductive inference has been successfully applied to image segmentation where the images are modelled by Markov random fields (MRF). We have extended this work to be capable of simultaneously reconstructing and segmenting images that have been observed only through noisy projections. The noise added to each projection depends on the classes of the pixels (material) that it passes through. The intended application is in lowdose (lowflux) Xray computed tomography (CT) where irregular projections are used. 1
Finding and using multiple . . .
 APPEARING IN THE 4 INTERNATIONAL CONFERENCE ON
"... Achieving autonomous learning systems which can govern themselves is one of the goals of A.I. Most learning systems explore a fixed model space to explain a set of data. We believe that the "best" but most distinct models in the available space can provide insight into questions of autonom ..."
Abstract
 Add to MetaCart
Achieving autonomous learning systems which can govern themselves is one of the goals of A.I. Most learning systems explore a fixed model space to explain a set of data. We believe that the "best" but most distinct models in the available space can provide insight into questions of autonomy such as when to change the model space and how to generate new data points (via experiments). We explore this idea by focusing on clustering problems where the initial data is known to be insufficient to find the true model. We propose a method to generate new data points via experiments. Our approach results in convergence to the true model using half as many additional data points than if they were randomly selected.
MML Analysis of Melody
, 1997
"... ii Abstract Appreciation of pattern is central to creativity. Given automatic generation of models, the Minimum Message Length (MML) approach can evaluate how well a model explains particular data, in terms of the message length it gives. This allows automatic discovery of pattern, and thus a method ..."
Abstract
 Add to MetaCart
(Show Context)
ii Abstract Appreciation of pattern is central to creativity. Given automatic generation of models, the Minimum Message Length (MML) approach can evaluate how well a model explains particular data, in terms of the message length it gives. This allows automatic discovery of pattern, and thus a method of simulating creativity. This thesis investigates this second phase, by evaluating models of music with respect to a body of nursery rhyme melodies. Music is the general area chosen because it is largely nonrepresentational. Nursery rhymes are selected because they are both simple and foundational of more complex music. This thesis builds on previous work both in method and content. The MML method is compared with entropy measurement, some models are fully explored, while others are suggested that are only possible within the MML framework. The models focus on different aspects of music: rhythm, pitch, melody and implied harmony. A very simple model of rhythm successfully explains the rhythms of the melodies examined. A natural development of this model was able to explain more complex rhythms, such as those found in the &quot;Hallelujah &quot; chorus of Handel's &quot;Messiah&quot;, where it gave a much shorter message length than LempelZivWelsh compression. These models were crossvalidated against many simple melodies. The typical layout of music notation reveals the same pattern explained by this model thus, we formally explain what every musician implicitly knows. These experiments also justify explicit reference to this layout when teaching music students to sightread. Another development of this model explains the simple phrase structure of a small corpus of iii nursery rhymes. The pitches of nursery rhymes were found to be better explained in terms of intervals than pitches, using zeroth order Markov models. A first order Markov model of intervals did no better than the zeroth order model. However, pitch was useful in describing the lowest note and the final note of a melody. Melody the interaction of pitch and rhythm was partially explained by literal and tonal repetition of bars, and by isolating runs of notes.
1 SNOB: A PROGRAM FOR DISCRIMINATING BETWEEN CLASSES
"... The problem addressed by Snob is known variously as "classification", "intrinsic classification " (to distinguish it from the case when the number and nature of the classes are known a priori), "clustering", "unsupervised pattern recogni ..."
Abstract
 Add to MetaCart
(Show Context)
The problem addressed by Snob is known variously as &quot;classification&quot;, &quot;intrinsic classification &quot; (to distinguish it from the case when the number and nature of the classes are known a priori), &quot;clustering&quot;, &quot;unsupervised pattern recognition&quot;, or &quot;numerical taxonomy&quot;. The decision statistic used by Snob to allocate objects to classes, to divide classes and to merge classes is called the Wallace Information Measure (WIM). Previous papers have presented the formal derivations of the Wallace Information Measure for the classification problem (Wallace &
GRADUAL DETERIORATION TRENDING AND FAULT DIAGNOSIS IN CUTTING TOOLS USING INDUCTIVE INFERENCE CLASSIFICATION
, 1993
"... AbstractAn effective procedure for the early detection and objective diagnosis of faults in turning machine cutting tools is described. The procedure involves the use of an inductive inference theorybased classification program called "Snob". The program objectively divides freq ..."
Abstract
 Add to MetaCart
(Show Context)
AbstractAn effective procedure for the early detection and objective diagnosis of faults in turning machine cutting tools is described. The procedure involves the use of an inductive inference theorybased classification program called &quot;Snob&quot;. The program objectively divides frequency spectra, generated from force measurements, into classes representing different cutting tool conditions. The estimated description length of each spectrum, which is used for classification, can also be used to detect the early stages of cutting tool deterioration. The procedure was tested using frequency spectra representing various stages of wear development in cutting tools during both accelerated wear rate and normal wear rate cutting tests. It is shown that the inductive inference theorybased spectra classification procedure allows early detection of cutting tool deterioration. 1.
Selection and Estimation of a Finite Generalized Dirichlet Mixture Model Based on Minimum Message Length
"... Abstract—We consider the problem of determining the structure of highdimensional data without prior knowledge of the number of clusters. Data are represented by a finite mixture model based on the generalized Dirichlet distribution. The generalized Dirichlet distribution has a more general covarian ..."
Abstract
 Add to MetaCart
(Show Context)
Abstract—We consider the problem of determining the structure of highdimensional data without prior knowledge of the number of clusters. Data are represented by a finite mixture model based on the generalized Dirichlet distribution. The generalized Dirichlet distribution has a more general covariance structure than the Dirichlet distribution and offers high flexibility and ease of use for the approximation of both symmetric and asymmetric distributions. This makes the generalized Dirichlet distribution more practical and useful. An important problem in mixture modeling is the determination of the number of clusters. Indeed, a mixture with too many or too few components may not be appropriate to approximate the true model. Here, we consider the application of the minimum message length (MML) principle to determine the number of clusters. The MML is derived so as to choose the number of clusters in the mixture model that best describes the data. A comparison with other selection criteria is performed. The validation involves synthetic data, real data clustering, and two interesting real applications: classification of Web pages, and texture database summarization for efficient retrieval.
International Journal of Electrical and Computer Engineering 2:7 2007 Support Vector Machine for Persian Font Recognition
"... Abstract—In this paper we examine the use of global texture analysis based approaches for the purpose of Persian font recognition in machineprinted document images. Most existing methods for font recognition make use of local typographical features and connected component analysis. However derivati ..."
Abstract
 Add to MetaCart
(Show Context)
Abstract—In this paper we examine the use of global texture analysis based approaches for the purpose of Persian font recognition in machineprinted document images. Most existing methods for font recognition make use of local typographical features and connected component analysis. However derivation of such features is not an easy task. Gabor filters are appropriate tools for texture analysis and are motivated by human visual system. Here we consider document images as textures and use Gabor filter responses for identifying the fonts. The method is content independent and involves no local feature analysis. Two different classifiers Weighted Euclidean Distance and SVM are used for the purpose of classification. Experiments on seven different type faces and four font styles show average accuracy of 85 % with WED and 82 % with SVM classifier over typefaces Keywords—Persian font recognition, support vector machine, gabor filter. I.