New Methods of Transforming Belief Functions to Pignistic Probability Functions in Evidence Theory

For many real time information fusion systems, one way to reduce the computational complexity is to establish safe decision threshold, and only those above the safe thresholds are considered in decision making. Pignistic probability transform is a useful tool for decision making by mapping belief functions to probabilities to improve decision credibility and reduce computational complexity as well. In practical systems, safe decision thresholds are often set in advance, so under the condition of not increasing the risk of wrong decisions, finding a reasonable probability transform to decrease the elements above the safe thresholds is essential. This paper introduces three new pignistic probability transforms based on multiple belief functions, then compares with other popular transform methods. Results show that the three methods are robust to mature or immature information sets, can decrease efficiently the elements above the safe decision thresholds, making the decision problem simpler.