Methods and systems for discriminating between objects of interest and background in images

Abstract

A method comprising: receiving a first image captured on an image sensor; obtaining a scene estimate for the first image, wherein the scene estimate comprises an estimate of the background signal for a region of space being imaged onto the sensor; storing the scene estimate in memory; receiving a plurality of successive images, wherein for each one of the successive images, the field of view is shifted with respect to the preceding image and overlaps partially with that of the preceding image; for each one of the successive images: retrieving the scene estimate from memory; subtracting the scene estimate from the image to obtain a background-subtracted image for output; updating the scene estimate based on the signal detected at the location on the image sensor to which the region of space is being imaged; and replacing the scene estimate stored in memory with the updated scene estimate.