An Automated Maritime Target Segmentation, Categorization and Parameter Extraction System

Automatic Target Recognition ATR is one of the intended capabilities of Integrated Submarine Imaging Systems (ISIS) to provide a submarine with the abilities of target categorization and distinguishing between hostile and non-hostile targets. For an ISIS system, ATR performance can be adversely affected by choppy sea, low visibility, water droplets on periscope head window, the presence of recticules in the image, pitch and roll of the submarine, and blurring due to periscope motion. Recent advances in computer vision and object recognition have brought new sets of local invariant features that have the potential to effectively address the aforementioned problems. Based on these achievements, UtopiaCompression Corporation (UC) proposes a MAritime Target Segmentation, CAtegorization and Parameter Extraction (MATSCAPE) system. This innovative ATR technology offers many advantageous capabilities over existing techniques, most notably including unexcelled accuracy, multi-level hierarchical classification, ability to recognize a target that is partially visible, extremely fast classification time, robustness to illumination changes and image transformations, ability to segment target outline, ability to extract target constituent components, adpativeness and incremental learning. The proposed technology aims to reduce operator workload and operator training time as well as to enhance submarine’s capabilities in intelligence, surveillance, reconnaissance and targeting missions.