关键词: 自动车牌识别轨迹数据, 伴随车辆组, DBSCAN, 豪斯多夫距离, 共现

Abstract: Automatic license plate identification(ANPR) s easier to obtain than private GPS data and it contains more useful information, but the relatively mature GPS track data mining with vehicle group method does not apply to ANPR data, the existing small number of automatic license plate identification data with the vehicle. In this paper a clustering method based on trajectory feature to excavate the vehicle group is proposed.The method is in the light of the character that the sampling location is fixed while the sampling time is stochastic in ANPR data, whether two objects form an adjoin model depends on the number of the co-occurrence of the trajectory. The co-occurrence definition introduces the Hausdorff distance, taking into account the location, direction and time characteristics of the trajectory. It is designed to mine the vehicle group with which have Similar trajectories but the sampling points are different and distance-closed, and this method improves the efficiency of mining with vehicle group. Experiments show that this method is more effectively excavate the vehicle group and identify the non-accompany model data than the existing method, and the efficiency has improved nearly twice.

Key words: ANPR trajectory data, Traveling companions, DBSCAN, Hausdorff distance, Co-occurrence