Human Communication Technology

10Hierarchical Elitism GSO Algorithm For Pattern Recognition

IlavazhagiBala S.¹ and Latha Parthiban²*

¹Bharathiar University, Coimbatore, India

²Department of Computer Science, Pondicherry University, CC, Pondicherry, India

Abstract

Medical imaging research has matured in the few decades as it is generally a non-invasive technique of diagnosis. Optimization of modular neural networks (MNN) is a requirement of gravitational search measurement and diagnosis during pattern recognition. Based on the analysis of conventional Gravitational Search Algorithms (GSA) for optimization of MNN in pattern recognition, a novel method called, Hierarchical Elitism Gene Gravitational Search (HEG-GS) is proposed. Here, echocardiogram videos are used as input both comprising of healthy and non-healthy patients. This paper concentrates an explicit representation of modules. First, the echocardiogram videos are split into frames and pre-processed using Additive Kuan Filter Pre-processing algorithm. By using this algorithm, the speckle noise present in the images is reduced. Second, with the pre-processed images, by applying a hierarchical elitism gene in GSA, the MNN architecture is optimized for pattern recognition. The performance of HEG-GS method is evaluated for echocardiogram pattern recognition task. Experiments with prototypic video from the Cardiac Motion and Imaging Planes validate that the HEG-GS method effectively performs pattern recognition and thus achieve improved recognition performance ...