Meta-optic Image Differentiation for High-speed Target Recognition

Fast and accurate object detection in the mid-wave infrared (MWIR) is critical for many DoD platforms including missile systems. Current object detection algorithms generally employ computationally heavy frontends to perform edge filtering tasks for keypoint identification and blob detection. Offloading the edge filtering tasks to optical components, which perform image differentiation or edge detection at the speed of light, offers the opportunity for significant latency reduction in these algorithms. In Phase I, the CFD Research team demonstrated the ability to design, fabricate, and test a metasurface optic (meta-optic) that successfully performs image differentiation of a scene. In Phase II, we will develop an optimized meta-optical system that performs multiple, parallel edge filtering computations in the optical domain to further reduce latency. We will employ end-to-end design methods that allow both the meta-optic and the digital backend to be co-optimized for improved, broadband performance. By the end of Phase II, we will demonstrate an advanced prototype that can be readily transitioned to relevant DoD programs in Phase III and beyond.