船舶管路所处的条件较为恶劣，发生泄漏难以避免。为了在管道泄漏初期能及时检测出微小的滴漏故障，进行维修避免造成更严重的泄漏，提出了一种管路滴漏视觉监测模型。模型中采用均值背景差分法检测管路滴漏，提取液滴前景的特征参数，结合虚拟线圈法，基于滴漏液滴个数进行流量统计，最后提出了滴漏体积流量的评估方案。为了验证模型的有效性，采取搭建滴漏实验台的方法获取滴漏视频对模型进行验证。结果表明：该模型能够很好的检测出管道滴漏液滴，特别是对于滴漏频次较慢的泄漏，能够准确的统计出滴漏个数且准确率在98%以上，泄漏体积流量的估计值相对误差在20%以内。此方法可以有效的对管路滴漏进行监测，为管路的维修决策的制定提供参考。

The conditions in which ship pipelines are located are relatively harsh, and leakage is difficult to avoid. In order to detect small drip faults in time in the early stage of pipeline leakage, and timely maintenance to avoid more serious leakage caused by negligence, a visual monitoring model of pipeline drip leakage is proposed. In this model, the mean background difference method is used to detect pipeline drip, output the characteristic parameters of droplet prospect, combine the virtual coil method to statistically count the number of drip droplets, and finally propose an evaluation scheme for drip volume flow. In order to verify the effectiveness of the model, the method of building a drip experiment bench was used to obtain the drip video to verify the model. The results show that the model can detect pipeline drip droplets well, especially for leaks with slow drip frequency, it can accurately count the number of drips, the flow accuracy is more than 98%, and the relative error of the estimated volume flow is within 20%. This method can effectively monitor pipeline drip leakage and provide reference for the making of pipeline maintenance decisions.