Moving object detection for video surveillance based on improved ViBe

In order to overcome the shortcoming of origin ViBe, such as background mutation caused by the changes of illumination and moving objects, and when the first frame contains a moving object, it needs a long time to eliminate Ghost Area. We present a novel moving object detection method based on improved ViBe. Firstly, the ViBe is used to detect moving objects and a background model Optimization Feedback Mechanism is proposed. Secondly, we establish a background update cycle strategy to accelerate the elimination of Ghost Area, combining with Inter-frame Difference to improve background mutation. Then a series of processing is made to optimize the detection results. Finally the improved object detection results are used as a feedback to background update process. Experimental results show that our method can extract the motion objects more accurately and the over-detection rate is low. It also has certain advantages in dealing with background mutation and noise suppression for video surveillance.