中文摘要：

用计算机来模拟雨景是影视制作中经常面临的一个课题。目前主要采用的粒子系统方法一般都需要大量的粒子和复杂的物理模型来模拟真实的雨景。提出了一种新的基于图像的方法来实现真实感雨景的绘制。用最小值滤波和改进的Matting算法相结合的方法,从视频中分割出真实的雨幕,然后根据目标场景的光照条件改变雨幕的颜色,最后将其与目标场景融合在一起。实验证明,该方法十分有效。同时该方法是完全基于视频图像的,避免了昂贵的3D场景建模,适合于低成本的影片制作和影视素材重用领域。

英文摘要：

Realistic rain rendering is a hot topic in the movie industry.It is usually simulated by particle system,which acquires a large number of particles with complex physical models to render photo-realistic rain.This paper proposes a new method to render photo-realistic rain by image based methods.This paper segments the realistic rain from video by minimal value filter and improves Matting algorithm.Then merge them into a new scene after translating their color style by different light condition.Experiments show the method is effective.Becanse of the method operates on the video image,it avoids the expensive 3D scene modeling.It is suitable for low cost movie maker and the video material reuse field.