基于自适应阈值的前景提取算法

Foreground Extraction Algorithm based on Adaptive Threshold

针对固定场景视频监控中运动目标提取的问题,提出了一种基于自适应阈值的前景提取方法。该算法通过混合高斯模型(GMM)对背景建模及更新,利用自适应阈值的方法,实现了模型门限的自适应调整和前景目标的分割。然后通过阴影抑制,滤波以及形态学处理的方法对前景目标进行后处理,改善了前景目标分割的质量。通过对不同场景的测试仿真表明,该算法能够有效地并且比较完整地提取出运动目标。

For the extraction of moving objects in the fixed scene surveillance,a foreground extraction approach based on adaptive threshold is proposed in this paper.The background model is established and updated through Gaussian mixture model（GMM）.The proposed algorithm,with the adaptive threshold method,succeeds in realizing the adaptive threshold adjustment and foreground object segmentation.Then,by such methods as shadow suppression,filtering and morphological processing,the quality of foreground object segmentation is effectively improved.Finally,simulation experiments on different scenes indicate that the proposed algorithm could implement effective and complete extraction of the moving target.