交通状态由畅通转入拥挤过程中和交通拥挤的时候容易发生车辆刮蹭、碰撞等交通事件,进一步减小道路的通行能力,加剧了交通拥堵.针对这种情况提出一种快速识别方法.通过统计累加方法获取并更新背景,并由背景差法获得前景点.在一个车道上设置不同区域获得时间占有率,并对一定时间内的前景点进行统计.当某一区域的前景信息与其他区域的前景信息不一样时,判断此区域内发生交通事故.实验表明所采用的方法能比较准确地获取前景信息,较好地检测出拥挤情况下的交通事件.%When traffic state transfers from smooth flow to congestion, vehicles scratch, collision and other traffic incident are prone to happen, which further reduce the passage capacity of roads, exacerbate traffic congestion, so for this situation a rapid identification method is proposed. The background through statistical accumulation method and gets foreground points through the background subtraction is obtained and updated. After setting different regions in one lane, the time occupancy and gather statistics of foreground points within a certain period are got. If the foreground information of a region is not the same as other regions, traffic incident happens in this region are determined. Experiments show that the method can be used obtain foreground information accurately and detect traffic incident well in congestion condition.
展开▼
