中文摘要：

针对现有PBAS目标检测算法在动态背景下存在着误检率高、检测精度较低的问题,提出了一种将改进的自适应决策阈值更新策略与优化处理方法相结合的目标检测算法。该算法首先使用改进的自适应前景判断阈值进行目标检测,然后对检测的结果使用前景点生命周期机制、形态学处理等方法进行优化处理。实验结果表明,与传统的PBAS算法相比,该算法在动态背景下可以更精确有效地提取出运动目标,准确度平均提高13%。

英文摘要：

As the result of the high misjudge rate of the existing PBAS algorithms and the low detection accuracy in dynamic scenes, an object detection algorithm with an improved adaptive decision threshold approach was proposed. Firstly, the improved adaptive threshold was used to determine the foreground and to detect the target to re- fine, Secondly, life cycle mechanism of foreground pixel and morphological processing etc were used on the result. Compared with the traditional PBAS algorithm, experimental results showed that the proposed algorithm can extract objects more accurately and efficiently in the dynamic background scenes, and the average of accuracy was improved by 13%.