自适应背景筛选的运动对象检测算法

针对传统帧差法和背景差分法对运动对象检测不准确等不足,提出了一种自适应背景筛选的运动对象检测算法。该算法在采用帧差法构建的背景中标注出原图存在运动对象的区域,筛选当前运动对象区域未被标注且距当前时刻最近的背景与当前帧进行差分,从而提取前景运动目标。与帧差背景结合方法相比,该方法能更好解决因运动对象静止后融入背景建模而导致的检测对象不准确问题,且算法简单,易于实现,满足实时监控要求。实验结果验证了该算法的有效性。

Moving objects detection algorithm with self-adaptive background screening

To improve the accuracy of the traditional coterminous frame differencing method and background subtraction method in moving objects detection.A moving objects detection algorithm with self-adaptive background screening is proposed.The area is marked,where the moving object exist in the graph,on the background in the process of constructing background model through the coterminous frame differencing method,.Background is screened that haven’t marked in the current area of moving object and the most close to the present moment,extract the foreground’s moving target.Compared with the traditional method which combine the coterminous frame differencing and background subtraction,the proposed algorithm is more good at deal with the matter,which detection is not accurate due to moving objects.It’s simple and easy to,satisfy the requirements in real-time monitoring.Experimental results verify the effectiveness of the algorithm.