Fusion of Radar and Electro-Optical/Infrared (EO/IR) for Ship Classification and Identification

Recent advances in machine learning are employed to provide substantial improvements in classification and identification by fusing information from multiple sensors. The envisioned approach uses deep learning networks to directly address apprehensions towards many machine learning applications related to the training dataset, the extent of the required real dataset for training and issues with reliance on simulations to generate training sets. Deep learning networks are used to generate valid, balanced and proper training sets to train the classification network processor. The proposed advanced machine learning approach will perform robust classification using ISAR and MWIR sensors with a path towards real-time, speedy classification performance. The Phase II effort will result in a prototype, a prototype demonstration and validation of the approach to meet the classification performance and real-time requirements.