Subjective Logic Based Approach to Modeling Default Reasoning for Visual Surveillance

In forensic analysis of visual surveillance data, `default reasoning´ can play an important role for deriving plausible semantic conclusions under incomplete and contradictory information about scenes. In this paper, we present an inference framework for default reasoning using Subjective Logic theory. Subjective Logic is a relatively new branch of probabilistic logic that allows explicit representation of ignorance about knowledge in a model called subjective opinion and that also comes with a rich set of operators thereby, having big potential as a tool for belief representation and reasoning. However, its application to visual surveillance is in its infancy and its use for default reasoning is not reported yet. Therefore, the aim of this paper is to bestow the ability of default reasoning on subjective logic and show the feasibility of using the introduced inference framework for visual surveillance. Among the approaches to enable default reasoning, the Bilattice framework is one that is well known and demonstrated for visual surveillance. For deriving the usage of subjective logic for default reasoning, we first discuss the similarity between the partial ignorance concept in subjective logic and the concept of degree of information in Bilattice based structure for multivalued default logic. Then we introduce the inference mechanism for default reasoning by mapping multi-logic-values into subjective opinion and combining operators in subjective logic. Finally, we present some illustrative reasoning examples in typical visual surveillance scenarios.