基于机器视觉的高速公路车辆停车检测

Vehicle parking detection on expressway using computer vision

为减少高速公路上车辆追尾等交通事故的发生,研究了基于机器视觉的车辆停车检测方法。检测系统采用当前帧和背景帧差分的方法实现了检测线圈范围内图像特征提取。分析了车辆从运动到静止的特征提取值图,并利用车辆存在阈值、车辆存在值、系统最小车辆存在值、系统不相似度阈值、系统相似度值、系统相似度最小值、车辆不存在值和系统最小车辆不存在值等参数实现了车辆停车检测。试验表明,该方法能够有效的将停靠在路面上的车辆检测出来。

In order to reduce the number of traffic accidents, such as rear-front collision, vehicle parking detection method based on computer vision is studied. Image feature within loop area is extracted by using image difference between current frame and background frame and image feature values are analyzed while moving vehicle starts to stop. Some parameters, such as vehicle existing threshold, vehicle existing value, minimum vehicle existing value, system dissimilarity threshold, system similarity value, minimum similarity value, non-vehicle existing value and minimum non-vehicle existing value, are introduced to realize vehicle parking detection. Experiment shows that this method can effectively detect vehicle parking on the road.