基于高斯模型的运动目标检测方法研究

提出了一种改进的基于单高斯背景模型运动目标的检测算法。该算法用四帧差分法确定运动目标区域,用单高斯模型对背景进行更新,然后通过(r,g,I)特征空间去除阴影,从而提取运动目标。实验结果表明,该算法能在变化的场景中对背景进行实时更新,能快速检测出运动目标,算法的运算量小。在室内环境和背景较为稳定的室外环境中都具有较好的应用价值。

Moving object detection based on improved Gaussian model

This paper proposes a moving target detection algorithm based on single Gaussian background model.The algorithm uses four frame difference method to determine the movement target area at first,and background model is updated by single Gaussian.Finally moving the shadow by(r,g,I)feature space.So the moving object is extracted.Experiment results show that the algorithm can update in real-time the background in changing scene and detect the moving object quickly.The calculation is simple and robust.This algorithm works well in the laboratory test and field application.