Research Of Motion Detection And Tracking Algorithm In Video Stream

Motion detection and tracking in video stream is always an important question for discussion in digital image processing and computer vision, the technology of catching and detecting the moving object in video sequence is widely used in mane situation. Presently the main technology of motion detection includes optical flow, background subtraction and differencing.The thesis mainly discusses about the basic principle of motion detection and tracking and related technology, the method adopted in the thesis and the designing of motion detection algorithm, lastly this thesis experimentize the algorithm proposed in the thesis and analyses the results of the algorithm.For the method of background subtraction, the thesis mainly used Mixture Gauss Model to establish a steady background, which can better adapt and update the disturbance in the background, then subtract the background from current video frame to get the moving object.For the symmetric difference method, the thesis improves on the ecumenical differencing method, which bases on neighboring three frames. This method can eliminate the revealable background disturbances brought by object movement and can better get the shape and location of the moving object. The method is fast and has little calculation.On the designing of algorithm, the thesis combines the two methods, which can makes up the shortage of each other and can also get a more precise location and information about the moving object. Experiments also indicate that the combination is effective.At the same time, on the explanation of related technologies, the thesis tries to model the technologies, and discuss them in the view of physics and mathematics. An advantage of doing so is that the analyses will be clearer.