一种融合雨滴检测算法的混合高斯模型

A Mixture Gaussian Model Fusion of the Raindrop Detection Algorithm

介绍了一种融合了雨滴检测算法的混合高斯模型,针对雨天气候下的场景,对雨中运动目标的检测有较好的准确性。首先建立自适应混合高斯模型,然后结合混合高斯模型参数对雨天视频图像进行雨滴检测,并恢复得到一张较贴近无雨状态的图像,再用混合高斯模型对运动目标进行检测,最终得到一个贴近真实目标的检测区域,并对混合高斯模型参数进行自适应更新。

In this paper,a mixture Gaussian model fusion of the raindrop detection algorithm is introduced,which is used for the scene of the rainy climate,and which can accurately detect the moving objection in the rain.First,Adaptive Gaussian mixture model is established,and using raindrop detection algorithm combined with the parameter of the Gaussian mixture model,then a image closer to the state of no rain is rebulided.Afterward,detecting the moving objection by use of the mixture Gaussian model,finally a detect region close to the true objection is obtained,and update the parameters of mixture Gaussian model.