一种基于轮廓和混合高斯模型的运动车辆视频检测方法

Approach for Moving Vehicle Video Detection Based on Contour Mixture Gaussian Models

针对光线突变对视频中运动目标检测的干扰问题,根据光线变化对事物轮廓没有直接影响的特点,提出了利用轮廓边缘元素建立自适应混合高斯背景模型的检测方法,有效地解决了由于光线突变造成的背景建模失真和前景目标难以准确提取的问题.实验表明该方法能够准确、实时地检测出复杂天气条件下的运动车辆目标,具有很强的鲁棒性.

Aiming at the sharp light changing interference problem of video detecting moving objects, according to the principle of the contour does not change with the light, we proposed an algorithm that using the contour edge image to build mixture Gaussian model. Experiments show that this method can real-time and accurately detect the moving vehicle in the complex weather, and demonstrate the robustness of it when the light suddenly had strong changes.