Moving of an object is studied as a continunous line in moving object database. So called trajectory that capture the convoy of moving objects becomes important in the trajectory databases.The discovery of convoy among database may be used for the scheduling of the identification of cars that follow the same routes, the using of the organization of carpooling, and the reducing pollution of CO2 emissions.Firstly, a new framework is proposed to discover convoys in the paper. The old algorithm incurs high computational cost because it performs clustering at every time and defines the actual tolerance as the maximum tolerance value over the all trajectory's line segments. We propose a new framework to get a better algorithm of convoy discovery to deal with these problems.Secondly, the paper defines the actual tolerance as a peremeter to tighten error bounds. As greater error bounds of trajectory will be reduced to a smaller one by cutting time .So it is easy to find convoys in the smaller time area. It won't perform clustering process at the timepoints that been cut.Finally, the paper proposes an idea of finding the trajectory fluctuation by variance, using the size of fluctuation to find convoys. As the cut of greater trajectory can effectively reducing the error bounds, we cut off the trajectory with large changes before and after the simplification. At the same time, we record the cut time to make sure that convoys are found in the same length of cut period times, ensuring the results are accurate and reducing the computing costs. The experiment demonstrates that the proposals are effective and efficient. |