Sourav De

Efficient grey-level image segmentation using an optimised MUSIG (OptiMUSIG) activation function

The conventional multilevel sigmoidal (MUSIG) activation function is efficient in segmenting multilevel images. The function uses equal and fixed class responses, thereby ignoring the heterogeneity of image information content. In this article, a novel approach for generating optimised class responses of the MUSIG activation function is proposed so that image content heterogeneity can be incorporated in the segmentation procedure. Four different types of objective function are used to measure the quality of the segmented images in the proposed genetic algorithm-based optimisation method. Results of segmentation of one synthetic and two real-life images by the proposed optimised MUSIG (OptiMUSIG) activation function with optimised class responses show better performances over the conventional MUSIG counterpart with equal and fixed responses. Comparative studies with the standard fuzzy c-means (FCM) algorithm, efficient in clustering of multidimensional data, also reveal better performances of the proposed function.

Color image segmentation using parallel OptiMUSIG activation function

Segmentation of the different feature based data in a dataset is a challenging proposition in the image processing community. There exist different techniques to solve this problem satisfactorily. A color image is an example of three-dimensional dataset and it consists of a collection of three primary color intensity features. In this article, we focus on the segmentation of true color test images, based on all possible combination of color intensity features. A multilevel sigmoidal (MUSIG) activation function that is applied in the self-organizing neural network architecture is quite efficient enough to segment multilevel gray level intensity images. The function uses equal and fixed class responses, ignoring the heterogeneity of image information content. The optimized version of MUSIG (OptiMUSIG) activation function for the self-organizing neural network architecture can be generated with the optimized class responses from the image content and can be used to effectively segment multilevel gray level intensity images as well. This article proposes a parallel version of the OptiMUSIG (ParaOptiMUSIG) activation function with the optimized class responses for the individual features with a parallel self-organizing neural network architecture to segment true color images. The optimized class responses are generated in parallel using a genetic algorithm based optimization technique. A standard objective function is applied to measure the quality of the segmented images in the proposed genetic algorithm-based optimization method. Results of segmentation of two real life true color images by the ParaOptiMUSIG activation function show better performances over those obtained with a conventional non-optimized MUSIG activation applied separately on the color gamut.

True color image segmentation by an optimized multilevel activation function

A novel neuro-fuzzy-genetic approach is presented in this article to segment a true color image into different color levels. A MUSIG activation function induces multiscaling capabilities in a parallel self organizing neural network (PSONN) architecture. The function however resorts to equal and fixed class responses, assuming the homogeneity of image information content. In the proposed approach, genetic algorithm has been used to generate optimized class responses of the MUSIG activation function. Subsequently, the color images are segmented by applying the resultant optimized multilevel sigmoidal (OptiMUSIG) activation function. Comparative results of segmentation of two real life true color images indicate better segmentation efficiency of the OptiMUSIG activation function over the standard MUSIG activation function.

OptiMUSIG: An Optimized Gray Level Image Segmentor

A multilevel sigmoidal (MUSIG) activation function is efficient in segmenting multilevel images. The function uses equal and fixed class responses, assuming the homogeneity of image information content. In this article, a novel approach for generating optimized class responses of the MUSIG activation function, is proposed. Three different types of objective function are used to measure the quality of the segmentation in the proposed genetic algorithm based optimization method. Results of segmentation of two real life images by the optimized MUSIG (OptiMUSIG) activation function with optimized class responses show better performances over the MUSIG activation function with equal and fixed responses.

Automatic magnetic resonance image segmentation by fuzzy intercluster hostility index based genetic algorithm

Graphical abstractDisplay Omitted Segmentation of magnetic resonance (MR) images plays an important role in the medical science or clinical research. In this article, an application of a genetic algorithm (GA) based segmentation algorithm is presented for automatic grouping of unlabeled pixels of the MR images into different homogeneous clusters. Before the segmentation, the information about the optimal number of segments as well as the underlying pixel distribution of an image is not required in this method. The centroid of different segments is demarcated as active/inactive centroid by the fuzzy intercluster hostility index. After that, the test images are segmented by the selected active centroids. The optimal number of segments and their respective centroids are determined by this method. A performance comparison is manifested between the fuzzy intercluster hostility index based GA method and the well-known automatic clustering using differential evolution (ACDE) algorithm and one genetic algorithm based non-automatic algorithm with the help of two real life MR images. The comparison depicted the superiority of the GA based automatic image segmentation method with the help of fuzzy intercluster hostility index over other two algorithms.

Multilevel color image segmentation using modified genetic algorithm (MfGA) inspired fuzzy c-means clustering

Convergence to local minima point is one of the major disadvantages of conventional fuzzy c-means (FCM). Due to this drawback, segmentation result may hamper for not selecting the cluster centroids properly. To overcome this, a modified genetic (MfGA) algorithm is proposed to improve the performance of FCM. The optimized class levels derived from the MfGA are employed as initial input to FCM for finding global optimal solutions in a large search space. An extensive performance comparison of the proposed MfGA inspired conventional FCM and GA based FCM on two multilevel color images establishes the superiority of the proposed approach.

Note: Measuring capacitance and inductance of a helical resonator and improving its quality factor by mutual inductance alteration

Narrow bandwidth and high voltage radio frequency (RF) is an essential requirement for stable confinement of ions within a RF trap and helical resonators are commonly used for that purpose. Effective capacitance and inductance of a helical resonator are estimated by measuring resonant frequencies for different external loads. Load capacitance of an ion trap can be estimated from this method and a resonator can be constructed for desired resonant frequency. We demonstrate a very simple method to achieve higher Q-factor of a resonator by optimizing mutual separation between the primary antenna and helical coil. We also formulate a set of analytical equations for calculating overall inductance, resistance, and Q-factor of a loaded helical resonator.

A differential evolution algorithm based automatic determination of optimal number of clusters validated by fuzzy intercluster hostility index

Automatic data clustering through determination of optimal number of clusters from the data content, is a challenging proposition. Lack of knowledge regarding the underlying data distribution poses constraints in proper determination of the inherent number of clusters.

Multilevel Image Segmentation Using OptiMUSIG Activation Function with Fixed and Variable Thresholding: A Comparative Study

An optimized multilevel sigmoidal (OptiMUSIG) activation function for segmentation of multilevel images is presented. The OptiMUSIG activation function is generated from the optimized class boundaries of input images. Results of application of the function with fixed and variable thresholding mechanisms are demonstrated on two real life images. The proposed OptiMUSIG activation function is found to outperform the conventionalMUSIG activation function using both fixed and variable thresholds.

Gray Scale Image Segmentation by NSGA-II Based OptiMUSIG Activation Function

A multilevel gray scale image can quite efficiently be segmented by the multilevel sigmoidal (MUSIG) activation function based on equal and fixed class responses, ignoring the heterogeneity of image information content. The optimized version of MUSIG (OptiMUSIG) activation function can be generated with the optimized class responses from the image content and can be used effectively to segment the multilevel gray scale images. These methods may or may not generate a good quality segmented image as the segmentation criteria of these methods are based on single segmentation evaluation criterion. This article proposed a self supervised image segmentation method by a non-dominated sorting genetic algorithm-II (NSGA-II) based optimized MUSIG (OptiMUSIG) activation function with a multi layer self organizing neural network (MLSONN) architecture to segment a multilevel gray scale intensity images. Some standard objective functions are applied in this proposed method to measure the quality of the segmented images. These functions form the multiple objective criteria of the NSGA-II based image segmentation method.