Automated Curvilinear Mineline Detection -- 19-011

Metron will develop an end to end processing chain of the COBRA imagery thru classification of minelike objects & identification of minelines to the User. During the initial phase of this effort, Metron will develop advanced processing algorithms for the detection of minelike echoes and then algorithms to fuse those echoes into groups of minelines. First, each of the six images in the spot will be processed including minelike detections. Then, the individual image detections will be fused at the spot level, creating a single set of detections for each proposed mine in a spot. Then area fusion brings detections from multiple spots into a single tactical picture. From here, minelines will be detected via curvilinear constrained or topography informed detection algorithms (the former detects a pattern that is uninterrupted, while the latter would detect using geographic information to inform placement, like a ridge). The minelines from each algorithm are fused into a set of minelines and then presented to the User. Prototype algorithms will be developed and delivered in Phase II and Metron will support transition into the COBRA Real Time Processor (RTP) framework as well as commercial applications in Phase III.