Real-time Hyperspectral Image Extraction from Plume Signature Databases based on Surrogate Modeling in Compressed States

The proposed effort aims to develop and demonstrate an innovative software tool combining surrogate modeling and image compression techniques for high-speed hyperspectral image extraction from plume signature databases. Salient aspects of the proposed solution are: (1) image compression and data-driven surrogate modeling operate in concert to capture the effects of various parameters (Mach number, angle of attack, altitude, etc.) on plume characteristics; (2) image compression, interpolation, and morphing/modification are fully coupled and performed concurrently in compressed images; (3) utilization of global statistical patterns across hyperspectral image sets, well-suited for database with sparsely placed nodes; and (4) a modular software environment to facilitate seamless integration into USAF-designated framework for sensors development and HWIL testing. In Phase I, key components, including a hyperspectral image compression module, a surrogate modeling module, a verification module, and an integrated software environment will be developed. Feasibility will be demonstrated via multiple case studies of USAF interest, in which synthetic and hyperspectral image sets at various flight parameters will be analyzed, and software performance (e.g., accuracy, speed, resource, and integrability) will be assessed. The Phase II effort will focus on capability extension, algorithm optimization, software integration, and extensive software validation and demonstration.