CLASSIFICATION OF SONAR IMAGES USING MULTILAYER PERCEPTRON NEURAL NETWORK

Abstract

In this paper another characterization calculation is proposed for the Classification of five sort of Side scan Sonar images. So as to create calculation 83 Side scan Sonar images have been viewed as, With a view to extricate highlights from the Side scan Sonar images after picture handling, a calculation proposes FFT changed coefficients. The Efficient classifiers in view of Multilayer Perceptron (MLP) Neural Network. A different CrossValidation data set is utilized for appropriate assessment of the proposed characterization calculation as for imperative execution measures, for example, MSE and order precision. The Average Classification Accuracy of MLP Neural Network containing one concealed layers with 13 PE's sorted out in a run of the mill topology is observed to be predominant (88.33 %) for Training and cross-approval. At long last, ideal calculation has been produced based on the best classifier execution. The calculation will give a powerful option in contrast to conventional strategy for Side output Sonar pictures examination for Classify the five kind of Side scan Sonar images in sea.