中文摘要：

通过背景差法对码头视频图像中的运动船只检测。采用混合高斯模型为码头背景建模,该模型可克服各种外部因素的影响,如光照变化、阴影、目标遮挡等,具有很强的适应能力。对采用全局固定阈值进行运动目标分割的方法作了优化,并对运动船只目标进行分割。经现场测试,能有效地监测运动船只、记录异常事件以及视频图像的提取和过滤,提高了码头监管的质量和效率,可广泛应用于码头无人监控。

英文摘要：

Background subtraction was used for detecting moving vessels in dock video images, Background modeling was based on mixture Gaussian model, which had strong adaptive capability for variety of external factors, such as illumination change, shadow, object shelter and etc, The moving vessels were segmented by an improved method using global fixed threshold. System testing in locale dock proved： the system could do moving vessels detecting, abnormal events registering and video abstraction and filtering effectively, which improved quality and efficiency, of dock inspecting and could be applied to dock control without man.