摘要
在基于视频的交通监控系统中,场景图像的分割在车辆检测中是一个重要环节,本文比较了传统的阈值分割方法,对一般迭代法提出改进,并引入一个平滑系数,使分割阈值在原来的水平上得到提高,并有效地避免当前帧分割阈值的突变。实践证明该算法能简单、鲁棒地实现分割,效果明显。
In vision- based traffic monitoring system, image segmentation is an important part in vehicle detection. This paper compared the traditional threshold segmentation, improved the conventional iterative algorithm, and introduced a smoothness coefficient, it can improve the threshold segmntation, and avoid the mutation in current frame. It is proved that the algorithm can realize segmentation simply and robustly , the effect is evident.
出处
《信息技术》
2005年第9期10-12,60,共4页
Information Technology
基金
上海市教委基金(217302)
关键词
交通监控
车辆检测
图像分割
traffic monitoring
vehicle examination
image segmentation
