Rapid Detection of Foreground Target Mutations Based on Risk Decision of Gaussian Mixture Background Model

The adaptive Gaussian Mixture Background Model is often used to detect moving target with fixed camera. In this paper， risk decision is introduced to judge mutations of foreground target. A timing counter function is designed to record the times of a pixel judged as foreground pixel. When the times is greater than a certain threshold， the pixel will be judged as a background pixel instead of a foreground pixel. So the probability of background pixel can be evaluated for making risk decision， updating the GMM parameters and reducing computation of the multiple Gaussian model. Finally， the algorithm is verified by experiments in the target detection rate and real-time improvements.