Detect-track-confirm filter with minimal constraints

We describe the theory of a detect-track-confirm filter whose role is moving target detection and clutter suppression in surveillance data. The filter has broad generality due to the minimal assumptions made in developing the theory. Track confirmation is decided on the basis of a probability measure that is fully computable from clutter properties measured from surveillance data, without needing to assume target properties such as trajectory or detectability. Experimental results on real surveillance datasets are presented.