摘要
针对视频图像中车辆检测问题,提出了一种基于计算机视觉的运动车辆检测方法。首先,选取最佳方差参数并利用双边滤波对图像进行预处理;再利用Surendra背景更新算法实时更新背景图像,采用粒子群(PSO)极大熵法求得背景差分图像的阈值;然后,将得到的二值图像进行形态学处理并检测出运动车辆。实验结果表明,该算法不仅适用于简单背景、车速较慢的环境中,而且在复杂背景、噪声较大、车速较快的情况下,该方法均能够克服外界环境的不利影响,准确地检测出运动车辆,提高检测的准确率。
A moving vehicle detection method was proposed, which is based on computer vision for vehicle detection in video image. First, the best variance parameters were selected and the bilateral filtering is used for image pre-processing. Then the real-time background image is updated with surendra background updating algorithm, using the particle swarm optimization (PSO) maximum entropy method to obtain background subtraction image threshold. Next, the binary image will be handled by morphology processing and the moving vehicles will be detected. Experimental results show that the algorithm is not only suitable for a simple background with slow-speed cars, but in a complex situation with larger noise and fast-speed cars, the proposed method can detect moving vehicles accurately through the adverse effect of the external environment. Hence, the accuracy of detection was improved.
出处
《计量学报》
CSCD
北大核心
2017年第3期288-291,共4页
Acta Metrologica Sinica
基金
国家自然科学基金(61601400)
河北省博士后基金(B2016003027)
关键词
计量学
运动车辆检测
粒子群算法
背景更新
视频监控
metrology
moving vehicle detection
particle swarm optimization algorithm
background updating
video surveillance