Target detection in M-disparate sonar platforms using multichannel hypothesis testing

The use of multiple disparate sonars allows one to exploit a high resolution sonar with good target definition while taking advantage of the clutter suppressing abilities of a low resolution broadband sonar co-registered over the same region to provide potentially much better detection and classification performance comparing to those of the single sonar cases. In this paper the standard Neyman-Pearson detector is extended to the dual disparate sonar case allowing target detection across two sensory platforms simultaneously. For three disparate sonar platforms, two detectors are implemented with the final decisions being fused. Test results of the proposed methods on a data set of underwater side-scan sonar imagery are presented. This database contains data from 3 different side-scan sonars, namely one high frequency sonar and two broadband sonars, operating at three different frequencies and bandwidths. The data was collected in different bottom conditions and contains various mine-like and non-mine-like objects with varying degree of difficultly and bottom clutter. Test results illustrate the effectiveness of the proposed detection system in terms of probability of detection, false alarm rate, and the receiver operating characteristic (ROC) curve.