Method for moving object detection based on adaptive filtration method and Bayes theorem application for changing estimation

We describe a new video surveillance method for moving objects detection. It combines adaptive filtration method together with algorithm for changing estimation based on Bayes theorem. At first adaptive algorithm localizes margins of moving object, and then algorithm corrects its form by Bayes theorem. In comparison to other approaches, proposed method is robust to noise, shadow, illumination variations and background repeated movements. Here moving detection is performed over adaptive scheme, therefore preliminary information on foreground and background is not need. We developed software based on proposed method and tested it on following situational video streams: street, office, store, fountain that confirmed its effectiveness. Developed method can be used for Internet video surveillance systems and other similar applications.