基于卡尔曼滤波算法的城市交叉路口车辆检测及分类

为了向相关部门提供更多的过往车辆信息以满足道路交通的需求,设计了一个基于卡尔曼滤波算法的城市交叉路口车辆检测及分类系统,用于对过往的车辆进行检测、计数和分类.首先采用背景差分法和卡尔曼滤波算法对在检测区的车辆进行检测和跟踪;然后使用经过检测、处理的被测车辆图像触发距其最近的相机进行图形分割;最后,通过LDA分类器对分段车辆的几何形状及外观特征进行正确地分类.所提系统的有效性在摄取的3400帧视频序列上得到了验证,实验结果表明,系统的检测率可达97.44％,正确分类率可达88.0％,与先进的方法相比,取得了更好的检测性能.

Vehicle Detection and Classification in Urban Intersection based on Kalman Filtering Algorithm

To provide in advance information to relevant authorities on the road traffic demand, an urban intersection vehicle detection and classification system based on kalman filter is designed to detect, count and classify the passed vehicles. Firstly, background subtraction and kalman filter algorithm are used to detect and track individual vehicles throughout the detection zone. Then, the detected vehicles blob-area is used to trigger the segmentation unit which in turn extracts the vehicle while at a point closest to the camera. Finally, both geometric and appearance features of the segmented vehicles are passed to the LDA classifier for proper categorisation. The effectiveness of proposed system has been verified on video sequence with 3,400 frames taken in by self. Experimental results show that the system achieved a high counting performance of 97.44% with corresponding classification rate of 88.0% and it has better detection performance comparing with the latest approaches.