改进的基于GMM的运动目标检测方法*

针对传统混合高斯背景建模(GMM)在一些复杂场景下未能有效地描述背景,提出了一种改进算法.该算法引入更新和消退控制因子改进参数更新模型,并定量约束运动目标停留时间,采用从时间域上过滤得到的快速变化的背景进行背景减除操作,最后在空间域上对检测结果进行数学形态学的处理.实验结果表明,该算法能够提高背景建立和形成速度,增强对背景扰动和光照变化的抗干扰能力,对固定摄像机场景下运动目标的检测具有良好的鲁棒性.

Improved moving objects detection method based on GMM

Arming at the limitations of being effectively reflecting the background by the traditional Gaussian mixture background modeling (GMM) in some complex situations, this paper proposed an improved algorithm. This algorithm added updating and regression controlling factor to improve the parameter model, and limited residence time of moving objects quantitatively. The detection results were formed by background subtraction method with the background which was better to adapt to a rapidly changing environment and established in the time domain. Finally, applied the mathematical morphology for image post-processing in the space domain. Experiment results show that this algorithm, which is robust to detecting moving object under a scene of the fixed camera, can accelerate the establishment and the formation of the background and improve the anti-interference ability of external disturbance and effect of illumination efficiently.