基于三帧差分和滑动平均背景的运动目标检测

Motion Object Detection Based on Three Frame Differencing and Sliding Average Background Method

摘要:

针对传统三帧差分法在前景提取中容易出现“空洞”现象,对传统三帧差分法作改进,采用逻辑“或”代替原来的逻辑“与”,但同时产生了运动目标被拉长,轮廓模糊等问题；为了有效解决该问题,将滑动平均法和改进的三帧差分相结合,然而滑动平均法初始阶段更新背景较慢,因此二者有效结合的关键是对滑动平均法作改进,提出一种自适应更新滑动平均系数加速背景更新的方法；最后将改进三帧差分提取的前景和改进滑动平均法提取的前景进行逻辑“与”运算得到综合前景图像；实验结果表明该算法在完整性,准确性以及实时性方面能满足实际要求。

Abstract:

In view of the traditional three frame difference method in the foreground extraction is easy to appear "hollow" phenomenon, make improvement on the traditional three frame difference method,use the logical "or" instead of the original logic "and", but at the same time produce a moving target is elongated and contour blurring problem. In order to effectively solve the problem,combined the sliding average background method with improved three frame difference,however in the initial phase the sliding average background to update the background is slow,therefore the key of effective combination of the two method is to improve the sliding average method,puts forward an adaptive update moving average coefficient of acceleration background updating method. Finally improved three frame difference extracted foreground and the foreground of improved sliding average method to logic "and" to get comprehensive foreground image. Experimental results show that the algorithm can quickly and completely and accurately detect moving targets and meet the needs of real-time detection.