摘要
针对背景固定的交通监控视频中的运动车辆检测问题,提出了一种改进的基于背景差算法的运动目标检测方法。该方法改进了混合高斯模型,对图像进行了平滑滤波预处理,并利用形态学滤波方法对二值化的前景图像进行后处理。该方法提高了背景模型的环境适应能力,能够很好地适应背景改变和光照等变化。同时,也改善了视觉效果,使前景检测误差值降低了14%,可为后续交通参数的提取提供更为精确可靠的图像数据信息。
An improved method of detecting moving vehicles monitored in traffic video surveillance based on background subtraction to improve detection effieieney of moving vehieles is proposed. Gaussian mixture model is improved and a smoothing filter is used to preprocess image in this paper. Finally, binary foreground images are postprocessed with morphology filters. The result of the experiment shows that this method can make the background model more adaptive to environment such as the change of background and the illumination. This method improves visual effect and reduces the foreground detection error by 14 percent, so it can provide more accurate and reliable image data information for the extraction of transportation parameters in the future.
出处
《成都信息工程学院学报》
2010年第4期355-360,共6页
Journal of Chengdu University of Information Technology
关键词
信号与信息处理
数字图像处理
车辆检测
背景差
混合高斯模型
形态学滤波
signal processing
digital image processing
vehicle detection
background subtraction
gaussian mixture model
morphology filters
