PDR: A Performance Evaluation Method for Foreground-Background Segmentation Algorithms

We introduce a performance evaluation methodology called Perturbation Detection Rate (PDR) analysis for measuring performance of foreground-background segmentation. It has some advantages over the commonly used Receiver Operation Characteristics (ROC) analysis. Specifically, it does not require foreground targets or knowledge of foreground distributions. It measures the sensitivity of a background subtraction algorithm in detecting possible low contrast targets against the background as a function of contrast, also depending on how well the model captures mixed (moving) background events. We compare four background subtraction algorithms using the methodology. The experimental results show how PDR is used to measure performance with respect to detection sensitivity in interesting low contrast regions.