Fused Unified Sensor Imaging and Optimization Network (FUSION)

We propose to explore multiple image pre-processing techniques applied to frames acquired with uncooled thermal long wavelength infrared (LWIR) cameras to enhance information available to soldiers. The collection of proposed software modules is called Fused Unified Sensor Imaging and Optimization Network (FUSION). FUSION will integrate multiple cutting-edge modules to achieve a comprehensive set of objectives. We will develop advanced image pre-processing techniques to enhance raw LWIR frames, including non-uniformity correction, noise reduction, frame-to-frame stabilization, and ego motion deblurring. The system will implement real-time sensor fusion of LWIR and visible/near-infrared (NIR) imagery, utilizing AI/ML algorithms for super-resolution and adaptive image enhancement. A closed-loop system for real-time optimization of camera parameters will ensure optimal performance across various operational conditions. Additionally, we will design a distributed aperture system for multi-sensor integration, enabling flexible and cost-effective system architectures. One FUSION hardware concept, assuming a helmet-mounted apparatus, could be as illustrated on the right. Raw frames collected with an uncooled LWIR camera in the field are of limited use to a soldier. To maximize usefulness, multiple processing steps must be applied in near real-time. Such cameras produce image artifacts resulting from a warm sensor, such as nonuniform self-emission. A camera that is not mechanically stabilized will have large frame-to-frame motion in typical field operation, and, depending on the exposure time, may also exhibit motion blurring. LWIR cameras have much poorer resolution than visible/NIR cameras, and they contain only greyscale information related to target object temperatures. To validate our approach, we will develop objective metrics and tools for algorithm performance evaluation.By achieving these objectives, FUSION will provide soldiers with clearer, more detailed thermal imagery, enhanced object detection and recognition capabilities, and improved overall situational awareness across various operational environments. This comprehensive approach offers a significant advancement in uncooled LWIR thermal imaging technology.