一种波形相似度的车辆检测融合算法

利用磁阻传感器可以实现停车位的车辆{有,无}状态检测,但是磁信号容易受到相邻车位停车干扰而导致误检,针对该问题设计了一种数据融合算法,将车辆磁信号建模为分段线性时间序列,提出了形态距离公式度量时间序列的相似度,计算相邻车位传感器磁信号的波形相似度实现车辆检测,在路面停车区域实地部署了50个磁阻传感器节点,通过长达5个多月的实验,结果表明该算法提高了车辆检测准确率.

A fusion algorithm for vehicle detection based on waveform similarity

Magnetic sensors have recently been proposed for parking spaces monitoring. The magnetic signal is easy to be disturbed by the vehicles parking on adjacent spaces,which will result in mistake of vehicle detection. To solve this problem,a data fusion algorithm is proposed.The vehicle magnetic signal is modeled as linear time series, and a distance formula to measure the similarity between time series is proposed. The vehicle detection is realized by comparing the similarity of magnetic signals sampled from the adjacent sensors. Fifty sensor nodes are deployed on the roadside parking spaces. By running the system for five months,we observed that the algorithm improves the accuracy of vehicle detection.