Advanced 3D Object Identification System

During the Phase I effort, OPTRA developed object detection, tracking, and identification algorithms and successfully tested these algorithms on computer-generated objects of various shapes and sizes and on sample real-world image sequences of a DeltaII booster separation. The algorithms currently identify six basic shapes (sphere, disk, cone, thin rectangle, tube, and cube) with better than 90% accuracy and estimate their size in three dimensions. Object shape and three dimensional size information can be determine using images from a single digital camera while two cameras are needed to provide three-dimensional trajectory and ground impact position information. While the Phase I showed the feasibility of the approach, work still needs to be done during the Phase II effort refining and expanding the detection, tracking, and identification algorithms to take into account the effect of noise, obscuration (smoke, clouds, etc), and viewing angle. Also during the Phase II, we plan develop a robust optical system using all commercial off-the-shelf (COTS) equipment and then test the system with several live outdoor tests.