Abstract
Moving object detection was implemented in dynamic background based on background difference compensation. Background differential can effectively segment the moving object in static background. But in moving video, the camera motion causes corresponding movement of the target and background, which makes the prospect moving object hard to separate from the background. In order to detect moving object, we can compensate the movement of the background and transfer the dynamic background to static. Moving object detection in static background image was implemented using a new weights updating method that the weights were updated during a certain period. This method based on classical Gaussian mixture model improved the efficiency of image segmentation greatly. Moving object detection in dynamic background was realized using background differential compensation. The global motion of the background was established according to the affined parameters model. The model parameters were estimated by feature points matching based on the search strategy. Invalid matching points were eliminated using the method of distance consistency. Backward mapping was used to get the motion parameters of the background. After compensation of the background with the global motion parameters, frame difference between the current frame and the background can detect moving objects effectively. Experiments were done on computer with the programming tools of VS2010 and MATLAB. Experimental results showed that the algorithm based on differential compensation was effective.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have