Event-based sensor multiple hypothesis tracker for space domain awareness

Event-based cameras are an emerging technology that measure changes in a field of view by recording data for each pixel only when a thresholded change in log-light intensity occurs. We leverage the low volume of data produced to develop a multiple hypothesis tracker (MHT) that processes time-series data as opposed to traditional image frames with the intent to augment Space Domain Awareness (SDA) tracking operations. We implement measurement rejection based on data attributes extracted from a training set of real event-based data. Our MHT with the data reduction hypothesis test implemented identifies satellite tracks in 92.1% of the test data sets. Addition of a simple noise filter, limiting the hypothesis test evaluation and subsequent track generation to higher frequency events, increases the processing speed by 69.5%. However, the noise filter rejects more real signal information prior to the hypothesis test which results in a decrease to 76.3% of the data sets producing satellite tracks. The initial success of the MHT warrants further development of this promising technique to enable event-based sensing SDA operations