基于背景更新的运动车辆检测算法研究

Background Updating Algorithm in Moving Vehicle Detection

在基于视觉的智能交通系统中,运动车辆的检测是最基础也是最关键的步骤。目前运动车辆检测中最常用的方法是背景差分法。该算法的关键在于背景图像的获取,由于背景图像的动态变化,为了有效的对车辆进行检测,需要对背景进行实时更新。因此,提出了一种新的基于像素点灰度值变化的自适应更新背景的算法,该算法在背景变化的情况下,能实时地修正或更新当前背景图像,再结合差分法与阈值化分割出完整的运动目标。通过实验证明了算法的有效性和实时性。

In video-based intelligent traffic system, vehicle detection is one of the most basic and valuable techniques. At present, the famous widely used in video-based vehicle detection is the background subtraction algorithm. The key to the algorithm is to get the background image, but the background often change, the background needs to be updated in real time in order to detect the vehicle effectively. Therefore, a new self-adaptive background updating algorithm is proposed which based on the changes of the pixel-gray value. In the changed background, background, this algorithm is mainly applied to the moving object detection, which can modify or update the current background model（CBM）in real time. Moreover, it can extract the moving objects completely by combining the background image difference and the threshold. The experiment proved that the algorithm is effective and real-time.