Research on background subtraction based on center-symmetric local binary patterns
Wu Jiansheng, Zu Xuliang · 2013
In traditional motion detection, the problem is that light changing and shadow are sensitive for background subtraction and it is difficult to achieve real-time detection due to high dimension and a large number of calculating when suing center-symmetric local binary patterns (CSLBP). A reasonable method of combining both method is proposed and used for motion detection. Using improved background subtraction based on center-symmetric local binary patterns, it solved the time efficiency problem by eliminating the need for handling shadows and high dimension. Experimental results show that detection results are improved obviously and it could meet the requirements of real-time detection.