基于改进混合高斯模型的前目标提取算法

Foreground object extraction via improved gaussian mixture model

前目标提取问题已经得到广泛的研究,而传统的混合高斯背模型提取的前目标含有许多误差点。针对这一缺点,本文对静态态背和动态背两种情况建立一种基于局部密度特征的改进的混合高斯模型,并设计相应的提取算法。实验结果表明,新算法可以精确地将前目标与背目标分离,得到的提取效果基本不含误差点,明显优于传统的混合高斯模型。

Foreground object extraction has been widely studied and has achieved great success,the results obtained by traditional Gaussian mixture model contain many error points.To tackle this issue,we propose a local density feature-based improved Gaussian mixture model and design the corresponding algorithm.Extensive experimental results demonstrate the superiority of the proposed method over the compared ones.Particularly,the error point can be removed by the proposed method.