Uncertainty Management In A Rule-Based Automatic Target Recognizer

Abstract

Automatic target recognition (ATR) is one of the most challenging tasks for a computer vision system.It involves the determination of objects in natural scenes in different weather conditions and in the presence of both active and passive countermeasures and battlefield contaminants. This high degree of variability introduces considerable uncertainty into the vision processes in an ATR. This mandates both a flexible control structure capable of adapting as conditions change and a method for managing the uncertainty to aggregate evidence.The desired flexibility can be achieved with a rule -based system in which the knowledge of the effects of scene content and ancillary information on algorithm choices and parameter values can be modeled and manipulated.In this paper, we describe such a system. The uncertainty is modelled by a combination of fuzzy set theory and Dempster-Shafer belief theory.Several variations of these methodologies within the rule -based structure are explored. The results are compared using sequences of forward looking infrared images.

Dempster-Shafer theory for sensor fusion in autonomous mobile robots

“…Another aspect is the incorporation of heuristic domain knowledge into the inference process in a natural way by enlarging belief. The enlargement function provides a more formal mechanism, following [27], than the generation of compatibility relations [14] and interpretation rules [20]. The quantification of domain knowledge still relies on the competence of the knowledge engineer; while this is no different than other implementations, either DS or Bayesian, it certainly calls for additional research.…”

“…Subsequent references to "probabilities" will be reserved for true statistical probabilities (i.e., Bayesian) while "belief function" will be used for the DS probabilities. DS theory has been considered for a variety of perceptual activities including sensor fusion [7], [8], [14], scene interpretation [25], object/target recognition [19], [20], and object verification [44]. The systems of [19], [20], [44] are particularly relevant because they view DS in terms of sensor characteristics.…”

“…No theoretical justification was given for the compatibility relations, although they are conceptually similar to the enlargement of FOD's through formal assumptions about the differences between FOD's. Keller and Hobson [20] in 1989 implemented interpretation rules for translating the belief in one FOD to another, again with no explicit theoretical justification. At about the same time, Hutchinson, Cromwell, and Kak [19] noted that for their application some FOD's could be considered refinements of another and transferred belief according to [47], permitting an implementation that would remain fully consistent with DS theory.…”

“…However, these single sensor perceptual systems have not been entirely successful for more demanding tasks in navigation [33], target or goal recognition [7], [14], [20], [32], [52], and general scene interpretation [8], [25]. This has limited the potential benefits of mobile robots for applications in space, defense, and manufacturing.…”

<title>Application of Dempster-Shafer theory to a novel control scheme for sensor fusion</title>

“…[ 10,11,29] are the two most common foundations for probabilistic techniques, although recent work has introduced new methods such as conductivity [32] and the tie statistic [13]. Both Bayesian and Dempster-Shafer theory can serve as the foundation for possibilistic or fuzzy techniques [8, 19,35]. Boolean logic has also been implemented in sensor fusion systems [4].…”