基于模糊匹配的多级车牌匹配技术

Multistep license plate matching technology based on fuzzy matching

在当前技术中,车牌之间的匹配一般采用的是精确匹配,其不足以满足旅行时间、OD调查等研究的需要。针对此问题,设计了一种基于模糊匹配的多级车牌匹配技术。该方法通过限定车牌经过上下游交叉路口的时间间隔、计算车牌之间的编辑距离进行车牌的模糊匹配,将匹配成功的车牌的旅行时间用四分位数法进行异常数据的剔除,并通过北京市某路段的实测数据进行验证。实验发现多级车牌匹配技术可以很好地提高车牌的匹配率,从而为基于车牌的其他研究提供更有效、可靠的数据。

Accurate matching is commonly used in current license plate matching,but it is not sufficient enough to meet the needs of research on travel time and OD survey. Aiming at the problems,a new multistep license plate matching technology based on fuzzy matching is proposed in this paper. This approach can find the matching license plates by limiting the time inter？val of the license plate from the upstream intersection to downstream intersection,and calculating Levenshtein distance between the plate and others. In the end,the quartile method is used to eliminate the abnormal data of travel time of the successfully？matched license plate. It was confirmed by the measured data of a road in Beijing. It is found in the experiment that the multi？step license plate matching technology can improve the matching rate of license plate,which can provide more effective and reliable data for other research based on the license plate.