基于混合高斯背景建模的交通视频运动目标检测

目前最常用也最有效的运动目标检测方法是背景减除法,其中背景提取是背景减除法的核心.传统的运动目标检测方法无法解决场景的光线突变、背景图像发生变化以及前景运动目标物体的阴影干扰等问题.针对交通视频中背景模型的实际情况,采用混合高斯分布对视频背景进行建模,将前一帧视频图像与所建立的当前背景图像进行相减,得到车辆在当前时刻的运动图像,并将所得图像进行形态学去噪处理.通过相关的仿真实验,证明了该方法能够比较准确地检测出前景运动车辆目标.

Moving Obj ect Detecting of Transportation Video Based on Mixture Gaussian Model Algorithm

The most common and effective approach in moving obj ect detection is background sub-traction,and the core of the approach is background modeling.Traditional detection methods can not overcome the problems such as illusion mutation,changing background images and target shadow influence.The background model with mixture Guassian distribution is constructed for the background of the transportation video.The current vehicle motion image is obtained by the subtraction of the previous frame video image and the background model.And the mathematical morphology image denoising processing gets more ideal vehicle detection rendering.The experi-mental results show that this method can detect the moving targets precisely.