Development of onboard image processing software for MONSTER on the HiZ-GUNDAM satellite

HiZ-GUNDAM is a future space mission designed to detect high-redshift gamma-ray bursts (GRBs) using both an X-ray telescope and a five-band visible/near-infrared telescope, MONSTER. Because the satellite’s real-time alert system cannot downlink raw images, image processing must be performed onboard. MONSTER detects roughly one thousand sources in the field of view, one of which is the GRB afterglow. Since transmitting information for all sources may be impossible, the system must identify only a few tens of afterglow candidates. To accomplish this, a lightweight decision-tree algorithm was developed that can run on the onboard computer. Using five-band photometric measurements, it narrows down the candidates to several tens. Using the constructed decision tree, when detector noise is ignored, the true afterglow at z > 6 is included among these selected candidates with nearly 100% probability. In this presentation, I describe the details of this onboard candidate-selection algorithm.