High-Speed Image Processing in Heavily Degraded Environments Using a Machine Learning Classifier

Driving ground vehicles in heavily degraded visual environments brings significant risk since course corrections cannot be made safely without visual cues. Infrared imagery can improve visibility in degraded visual environments, however when heavily degraded, this method alone is insufficient. Noise sources can be rapidly categorized and suppressed in near real-time by applying modern image processing techniques with machine-learning-based classification strategies. Noise sources will first be analyzed in degraded imagery including the temporal component of the noise. A supervised machine learning approach will categorize these noise sources, allowing for rapid identification of the optimal noise removal method. These noise signatures will be next replicated, and then transforms developed that can repair damage done by such noise sources. Finally, a limited scope implementation will be done to ensure high-speed application of these image enhancements is suitable while driving a vehicle. Phase II efforts will fully implement the method and target below 80 ms latency. Phase III efforts will focus on widespread implementation.