利用FCM对静态图像进行交通状态识别

发布时间：2018-07-12 20:17

  本文选题：交通状态识别 + 交通图像　； 参考：《西安电子科技大学学报》2017年06期

【摘要】：对交通状态进行准确识别可以主动预警将要进入本路段的驾驶员避开拥堵,以免加重拥堵程度,同时也是科学制定主动交通管理决策的基础,有利于及时疏导拥堵,提高道路运行效率,节能减排.首先从交通监控视频中采集图像,标注道路为兴趣区,并对道路图像做角度和尺度的归一化处理;然后提取兴趣区图像的平均梯度、角点个数和长边缘比例3个特征;最后,利用模糊C均值聚类算法将图片所呈现的交通状态分为畅通和拥堵两种状态.实验结果表明,文中算法可以有效识别图像中的交通状态,正确率达到了94%以上,而且较基于视频的交通状态识别方法,该方法也大大降低了实现成本.
[Abstract]:Accurate identification of traffic conditions can actively warn drivers who will enter this section of the road to avoid congestion, so as to avoid aggravating congestion, and it is also the basis of scientific decision-making on active traffic management, which is conducive to timely dredging congestion. Improve road operation efficiency, energy conservation and emission reduction. Firstly, the images are collected from the traffic surveillance video, the road is marked as the area of interest, and the road image is normalized in angle and scale. Then, the average gradient, the number of corner points and the proportion of long edges are extracted. Finally, the average gradient of the image, the number of corner points and the proportion of long edges are extracted. A fuzzy C-means clustering algorithm is used to classify the traffic state presented by the image into two states: unblocked and congested. The experimental results show that the proposed algorithm can effectively recognize the traffic state in the image, and the correct rate is over 94%, which is much lower than that of the video-based traffic state recognition method.
【作者单位】： 长安大学信息工程学院;
【基金】：国家自然科学基金资助项目(61572083) 陕西省自然科学基金资助项目(2015JQ6230) 中央高校基本科研业务费专项资金资助项目(310824152009)
【分类号】：TP391.41
，

本文编号：2118345