784141.fig.006a
784141.fig.006b
784141.fig.006c
Figure 6: Motion hypotheses formulation. Three stages of a detection and classification of a transient. From left to right: the initial detection of the event on a first image, formation of a circular region of expectation based on assumption about some reasonable maximal speed of expectation; the formation of a set of hypotheses on object motion direction and speed based on an appearance of events inside this region on a second frame; the rejection of failed and acceptance of successful hypotheses based on events seen on a third frame. Objects not detected on a second frame are immediately rejected as a probably noise events; objects with determined motion parameters are tracked until their disappearance from field of view or fading out below the detection limit. For FAVOR and TORTORA systems with 0.13 s temporal resolution, this three-stage algorithm is completed in less than 0.4 seconds, resulting in nearly real-time detection and classification of transient.