Target detection enhancements using in-situ environment adaptive clutter modeling

The significant challenge in shallow water active sonar is to provide reliable target detection and classification while maintaining a reasonable false alert rate. When reverberation appears target-like due to persistence in location from ping to ping and exhibits a coherent intensity, an extremely high false alert rate occurs. This phenomenon occurs frequently in littorals where boundary interaction and bathymetry effects dominate the reverberation profile. In current systems the assumptions of the noise statistics are based upon Gaussian derived processes. The distribution of the envelope at the output of the matched filter is assumed to be Rayleigh. In conditions of high clutter the distribution deviates significantly from Rayleigh, where the tail of the density function tends to be significantly higher than the theoretical noise assumption. These conditions can be predicted and certainly can be measured by estimating the distribution of data at the detector output to adapt the detection processing based upon changes sensed in the background distribution. Performed in a spatial/range dependent manner environmental sensing can greatly reduce the clutter on a display and aid the automated detection process. A target detection technique based on the Interacting Multiple Model estimation algorithm is presented to exploit the consistency of position and motion information in conjunction with accurate models for target and clutter amplitudes observed over multiple pings to form a detection decision. The density and nature of the amplitude distribution for clutter, which is highly dependent upon the environment, is shown to dramatically affect detection performance. Performance improvements can be obtained by adapting the statistical model for reverberation based upon localized in-situ measurements to reduce the adverse effects of high amplitude clutter on multiple ping detection and classification. Using this knowledge an adaptive sequential likelihood ratio test that minimizes the decision time and improves detection performance under a vast set of environmental conditions is developed and demonstrated using collected sea test data