Cascaded Volterra fusion of processing strings for 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 overall CAD/CAC processing string consists of pre-processing, adaptive clutter filtering (ACF), normalization, detection, feature extraction, optimal subset feature selection, feature orthogonalization, classification, and fusion processing blocks. The range-dimension ACF is matched both to average highlight and shadow information, while also adaptively suppressing background clutter. For each detected object, features are extracted and processed through an orthogonalization transformation, enabling an efficient application of the optimal log-likelihood-ratio-test (LLRT) classification rule, in the orthogonal feature space domain. The classified objects of 4 distinct processing strings are fused using the classification confidence values as features and either "M-out-of-N" or LLRT-based fusion rules. The utility of the overall processing strings and their fusion was demonstrated with new shallow water high-resolution sonar imagery data. The processing string detection and classification parameters were tuned and the string classification performance was optimized, by appropriately selecting a subset of the original feature set. Two significant fusion algorithm improvements were made. First, a new nonlinear (Volterra) feature LLRT fusion algorithm was developed. Second, a repeated application of the subset feature selection/feature orthogonalization/Volterra feature LLRT fusion block was utilized. It was shown that this cascaded Volterra feature LLRT fusion of the CAD/CAC processing strings outperforms the "M-out-of-N" and baseline LLRT algorithms, yielding significant improvements over the best single CAD/CAC processing string results, and providing the capability to correctly call all mine targets while maintaining a very low false alarm rate