基于混合高斯背景模型的水面目标检测

通过混合高斯背景模型法对水利监控视频图像中的水面运动目标进行检测。根据河流水面的特点,采用混合高斯模型为水面背景建模,并运用在线EM算法进行背景模型的快速更新,然后用学习的高斯背景模型进行水面目标检测。该背景模型能够克服各种外部因素的影响,如光照变化、水面波浪,水面反光等,具有很强的适应能力,并能够满足实际应用中对目标检测的实时性要求。

Water Surface Target Detection Based on Gaussian Mixture Background Model

This paper introduced a study on water surface moving target detection of surveillance video images by method of Gaussian mixture background model. According to the characteristics of river water, adopted gaussian mixture models for water surface background modeling, and used an online EM algorithm to update the background model fast, then carried on the surface target examination with the study Gauss background model. This background model can overcome the influence of various external factors, such as the change of illumination, the wave on water surface, the reflection and so on, have a strong ability of adoptions, and can meet the requirements of the practical application of real-time target detection.