Comparative Study of Multiclass Classifiers for Underwater Target Classification

Abstract

Underwater target classification is a complex task, due to the difficulty in identifying non-overlapping and stable feature set. Moreover, choosing the discriminating algorithm for classification using these features is highly demanding. It is required to choose the right approach and the technique, or the best combinations of techniques from a large set of options available in the literature for the specific problem. The paper addresses this issue by comparing different approaches and techniques for multiclass classification using a particular feature derived from the real data sets. A number of performance metrices are used to compare the performance.