A background reconstruction algorithm for outdoor video surveillance

The accuracy of motion objects detection is low for outdoor complicated visual surveillance, so a background reconstruction algorithm used in background subtraction motion objects detection is proposed. The algorithm integrates the stabilization duration with appearance frequency, and uses weighted intensity histogram to confirm the background probability of quantity interval. Based on the probability values and distributing, it uses the intensity space correlation to amend the background, and realizes the background reconstruction. The experimental results show that the reconstructed background conquers the influence of motion objects in training phase, and in motion objects detection application, it is applicable in the situation existing background fluctuation and small camera displacements etc, and it improves the accuracy of detection.