Research on an improved visual background extraction algorithm

The visual background extraction algorithm (ViBe) uses the first frame of the video as the initial frame to initialize the background model. If the initial frame of the background model has a moving object or the state of the moving object changes (from still to motion or from motion to still) or the update rate of the background model is not synchronized with the background variability, there will be false target which is known as ghost. Aiming at this problem, a discriminant method based on histogram similarity measure of foreground region and neighborhood background region is proposed. The algorithm can detect ghost regions from the moving area. The experimental results show that compared with the traditional algorithm, the improved algorithm can effectively remove the ghost of foreground detection, and the detection of the object is more accurate.