Concepts for the Use of Three-Dimensional Depth Maps in Underwater MineClassification

摘要

This report is an examination of concepts for the utilization of threedimensional depth map information for the construction of features to be employed by neural network based classifiers in the classification of underwater mines. The evaluation of the discriminatory power of a subset of the identified three dimensional features for mine detection/classification, in the context of actual or simulated data, will be reported separately. The context that forms the basis for the current study envisions the availability of coregistered side-scan sonar derived intensity imagery and three dimensional depth map information at the same resolution. Hence, the three-dimensional related features to be described are to be viewed as augmenting features computed from side scan sonar imagery. The discussion focuses separately in Section 2 and 3 on the construction of three-dimensional features that use low resolution and high resolution depth map construction concepts, respectively, and associated with the use of telesounding and swath bathymetry techniques, respectively. Finally, Section 4 summarizes the study's conclusions.