基于统计信息的改进滑动平均目标检测算法

Improved sliding average target detection algorithm based on statistical information

随着实时视频监控技术的发展,运动目标的检测在众多领域都有着广泛的应用。针对当下运动目标检测算法易受环境噪声和光照突变的影响,为了提高算法的稳定性和自主检测能力,本文系统研究了背景减除法和帧差法的不足和优势,提出了一种基于统计信息的改进滑动平均算法。首先,依据帧间差分对光照不敏的特性,粗略分离出背景和前景部分,同时去除掉其他随机孤立噪声,然后在滑动平均算法中引入对前景点的统计计数,自适应更新背景建模学习速率,加速背景模型收敛,最后将滑动平均与三帧差分两者形态学处理后的目标团块进行与运算,得到准确的前景区域。实验结果表明,本算法可以在树叶晃动和光照变化中精确定位运动目标,正确前景点率平均提升7.4%,得益于帧差法特性,误检前景点率显著降低,鲁棒性和自适应性更好。

With the development of real-time video surveillance,moving target detection algorithm has been used in many fields.The current moving object detection algorithm is most vulnerable to environmental noise and light mutation.In order to improve the stability and self-adaptability of the algorithm,an improved moving average target detection algorithm based on statistical information is proposed.Firstly,on the basis of the characteristics of inter-frame difference that is insensitive about light,the background and foreground parts are separated and other random noise is also removed.Secondly,the information of statistical counting of foreground points is introduced in the moving average algorithm to adaptive update background modeling learning rate which is effective to accelerate the convergence of the background model.Finally,to get accurate moving target position we shouldprocess the results of the two algorithms by further using morphological methods and AND operation.Experimental results show that the proposed algorithm can accurately locate moving targets under the condition of light mutation and leaves shaking.Compared to the previous algorithm,the TP rate increases by 7.4% and the FP rate significantly decreases.The proposed algorithm has better robustness and adaptability.