F-22 Data Compression

The F-22 Data Compression program addresses the problem of the insufficient imagery storage capacity of the F-22. This impacts the ability of the aircraft to deploy munitions or transfer imagery for tactical planning. The imagery capacity of the targeting system is significantly smaller than the area of operations thus restricting the ability to attack targets. For reconnaissance operations, due to the large size of the images, the amount of data that can be recorded is limited and must be manually processed and distributed. The data compression algorithms used by the DRA/UtopiaCompression team will allow the F-22 SAR imagery to be compressed using a pattern driven approach whereby images and, specifically, regions in images with defined visual characteristics are considered as relational entities, classifiable according to the presence or absence of patterns, their relationships, and the frequency of their occurrence. The algorithm seeks to minimize the set of extracted features with the application of statistical machine learning to extract features and model patterns. The codec is trained on such patterns to “learn” their nature so that variable-sized segments of the image can be accurately reconstructed. The algorithm results in a five to ten fold improvement over JPEG compression technology.