混合交通流视频检测中区域选择更新混合高斯背景模型研究

Research Region Selective Update Mixture Gaussian Background Model In Hybrid Traffic Flow Video Detection Process

混合交通是我国交通的主要特征，利用视频检测技术可以获取混合交通流参数，实现混合交通的有效管制，由于检测过程中天气、光线等环境变化，实时有效的自适应背景提取模型尤其重要。本文在混合高斯模型的基础上，根据运动分割与Kalman运动跟踪，结合象素的时间与空间特性，提出区域选择更新混合高斯模型来抽取背景，克服了交通控制信号或交通阻塞等造成的长时间停车，以及高峰期大量运动物体长期充满当前图像等情况对背景抽取造成的影响，该模型通过对交叉口和路段视频进行背景提取，实验效果良好，证明了本方法具有较强的鲁椿性和自适应性。

The hybrid traffic is main property of china traffic. It can be obtained hybrid traffic flow parameter by video detection technology to achieve hybrid transportation management and control. For environment vary with time, such as weather, light and so on, it is crucial to use self-adaptive background model to extract static background. In the paper, To take into account temporal and spatial relation on pixels, It presents Region Selective Update Mixture Gaussian Background Model. After image segmentation and kalman motion tracking, the model updated by Region Selective and overcome the influence on vehicles stop caused by traffic control sign or traffic block for a long time and moving vehicles occupyed current image during traffic fastigium. In experiment, the video on crossway and a section of a highway were tested, and the experiment results show the update model is effectively, Consequently, It proved the method have better robust and self-adaptability.