Image sequences are used in various fields, including medical imaging and satellite/airborne imaging. The comparison between images taken at different conditions (e.g. equipment or configuration, angles, weather and wavelength) can be a highly non-trivial problem, as subtraction artifacts can outnumber real changes between images. The existing remedy for this problem includes highly complex solutions using machine learning algorithms to narrow the sea of candidates. In some cases, human interpretation of images cannot be avoided, resulting is very long processing times. The new method presented here provides a proven solution for the subtraction of images taken at varying conditions. The tool can be applied for any type of imaging, allowing fast processing and accurate results.
The new method is used for processing at least two N-dimensional data-measurements (DMs) of a physical-property for detecting one or more new-objects and/or a transition of one or more known-objects, in complex constant-background DMs. Generally, the the method includes: (1) generating a filtered-new-DM by match-filtering a new-DM, respective to impulse response of a reference-DM (2) generating a filtered-reference-DM by match-filtering the reference-DM, respective to impulse response of the new-DM (3) generating an N-dimensional object-indicator (OI) by subtracting the filtered-reference-DM from the filtered-new-DM, or vice versa and (4) generating an N-dimensional data score from the N-dimensional OI, where each of the scores is a probe for existence of an object at a specific N-dimensional location.