一种改进反馈机制的PBAS运动目标检测算法

摘要

To improve the adaptability of PBAS algorithm to various dynamic background , this paper proposed a variant which uses adaptive multi-channel background modeling and feedback of local segmentation noise levels. The proposed algorithm takes information of multi-channel into consideration and adjust each channel′s decision threshold and weight adaptively. The proposed algorithm uses statistical information of segmentation noise to adjust parameters without addition of new features , which made error rate further less. Experimental results showed that the proposed algorithm has a much better adaptability when used in dynamic backgrounds and it can provide a more accurate segmentation.%为了提高PBAS算法在不同动态背景场景时的适应能力,本文在原始PBAS算法的基础上提出了一种自适应多通道建模与分割闪烁点反馈的运动目标检测方法.该算法综合考虑多通道的信息,自适应调节各个通道的判别阈值以及权重;在不增加新特征的情况下,利用分割闪烁点统计信息反馈调整算法参数,减少错误判别结果.实验结果表明,该算法在动态背景场景下具有原PBAS更好的适应能力,在大多数应用场景能提高更好的分割结果.