Comparison of Volterra and Box-Cox Methodologies for the Fusion of Processing Strings, as Applied to Automated Sea Mine Classification in Shallow Water

An improved sea mine computer-aided-detection/ computer-aided-classification (CAD/CAC) processing string has been developed. The classified objects of 3 distinct strings are fused using the classification confidence values and their expansions as features and Fisher ratio based fusion rules. The utility of the overall processing strings and their fusion was demonstrated with new shallow water high-resolution sonar imagery data. Two significant fusion algorithm improvements were made. First, a new nonlinear 2
^nd
order expansion (Volterra) feature fusion algorithm was developed, and an optimal subset of the Volterra features was selected during the training phase of the algorithm. Second, another new nonlinear Box-Cox expansion of the features (raising the features to an appropriately selected exponent) was utilized in the fusion block. It was shown that, when properly formulated, the Box-Cox nonlinear feature fusion of the CAD/CAC processing strings outperforms the Volterra feature fusion algorithm, and also yields an improvement over the best single CAD/CAC processing string, providing a significant reduction in the false alarm rate