Research On The Double Cameras Video Surveillance Alarming Algorithm Based On Blob

Swimming has become one of the most popular sports, and the potential dangers have caused seriously attention. To only depend on the lifeguards can not make save to the swimmers. Therefore, we need the video surveillance system to help them. Because of single camera's view area is narrow and it also has other disadvantages, the double camera video surveillance system is eagerly required. This has been caused widely concerned both at home and abroad. According to the requires of the system, this paper combines with the algorithm of video image, makes detection and tracking to moving swimmers, and makes analysis to their behaviors.Depending on the Blob algorithm, we make clearly analysis to the six modules of it. We also introduce two FG detection methods, filter algorithm, CC method, Mean Shift, particle filter and Kalman filter. In different modules, we select different methods in experiments. At last, the best-fit methods are applied in detecting and tracking moving swimmer.According to the significant change of normal and drawing swimmers, referring to many statistics both at home and abroad, we summarize the drawing swimmers' significant features, such as center, velocity and shape. And we also make rules of them. With the state transfer diagram, we could determine whether a swimmer is in danger or not and make alarm.Obviously, the two cameras need to be matched, especially to the object. First, we judges if the moving swimming is in two cameras, that is one part in camera 1 and the other part in camera 2. We also make different methods for different features match, such as axis, shape and velocity, and makes further judgment with the whole swimmer. At last, we could determine weather the two parts belong to the same moving object and monitor it.Finally, this method could distinguish between normal and drawing swimmers effectively by experiments. And it also could alarm fast and accurately.