Abstract: In view of the limitation of the existing methods, a new algorithm of background extraction and updating was proposed. The interval pixel value belonging to the video stream was calculated to extract background. The edge detection method was referred to analyze the area of continuous change when there was stagnant vehicle in the current background image. The testing results demonstrate that the proposed method can be well adapted to the application of the complicated situation with markedly improved accuracy, and it is simple and suitable for real-time detection.

Key words: background extraction, background updating, interval distribution, edge detection, vehicle detection

摘要： 现有算法对长时间含有静止车辆的视频序列进行背景提取和更新时效果不甚理想，为此提出一种改进的道路背景提取和更新算法。通过统计一段时间内视频帧中各像素值所属区间来提取背景图像，结合边缘检测算法分析持续变化的区域，解决背景更新中存在的当前背景图像中含有静止车辆的问题。实验表明算法可以很好地适应实际应用中的复杂情况，准确性有了明显提高,且算法实现简单，实时性较高。

关键词: 背景提取, 背景更新, 区间划分, 边缘检测, 车辆检测