结合三帧差分法和混合高斯模型的运动目标检测

鉴于传统的三帧差分法(TFDM)在检测结果上有检测目标不完整的问题,提出了一种将TFDM与混合高斯模型(GMM)相结合的运动目标检测方法(TFDMGMM算法)。该方法用多个GMM的加权表示每一个背景像素,再结合TFDM判别是前景还是背景。实验结果表明,与单独的TFDM和GMM算法相比,既能得到相对完整的运动目标,又能减小噪声等外部干扰带来的影响。

Motion Target Detection in Combination with Three-Frame Differential Method and Gaussian Mixture Models

In view of the problem that the traditional three-frame difference method has incomplete detected targets in the detection results, this paper proposes a moving target detection method which combines the mixed Gaussian model with the three-frame difference.In this method, the superposition of multiple mixed Gaussian models is used to represent each background pixel, and the three-frame difference method is used to distinguish the foreground from the background.The experimental results show that compared with the three frame difference method and the mixed Gaussian algorithm, it can not only get a relatively complete moving target, but also reduce the influence of external interference such as noise.