一种改进的自适应高斯混合模型实时运动目标检测算法

Improved algorithm of adaptive Gaussian mixture model for real-time moving object detection

高斯混合模型被广泛应用于摄像机静止条件下运动目标检测的背景建模。针对传统高斯混合模型中对光照变化适应性差及学习率单一等问题,提出了一种光照变化检测及学习率更新的方法,以达到自适应更新背景模型的目的。提出利用颜色直方图匹配算法,通过引入光照变化因子以及模型参数更新计数器对学习率进行自适应的调整,并通过对描述模型分量个数的自适应选择减少了计算时间,增强了系统的实时性。实验结果表明,该方法能快速有效地适应场景的变化,比传统高斯混合模型具有更好的鲁棒性与稳定性。

Gaussian mixture model is a widely used approach for background modeling to detect moving objects from static cameras. Base on the situation that traditional Gaussian mixture model could not effictively solve the problems such as poor adaptability of illumination change and single of learning rate, this paper proposed a method to detect the illumination variation and update the learning rate, in order to achieve the purpose of the adaptive updating background model. Firstly, the proposed method used the color histogram matching algorithm to introduce the factor of illumination variation. At the same time, it also introduced a counter of the updating parameters of model to adaptively adjust learning rates. And then, adaptive selection of the number of components in the model reduced the computation time and enhanced the real-time nature of the system. The experimental results show that this proposed method can quickly and effectively adapt to changes in the scene, and has better robustness and stability than traditional Gaussian mixture model.