基於移動軌跡建模之異常軌跡偵測方法

Abstract

全球定位系統（Global Positioning System，GPS）幾乎已成為生活中的一種重要工具，在各種交通工具皆可見到全球定位系統的應用。全球定位系統不只提供定位及導航功能，更可以利用其提供之軌跡資料進行異常軌跡的偵測。在本論文中，我們實作一個基於移動軌跡模型之異常軌跡偵測方法，利用歷史軌跡建立軌跡樣式樹模型，當移動軌跡建模完成後，我們就可以將即時軌跡輸入軌跡樣式樹模型中來偵測是否存在異常軌跡。最後敘述了實驗資料及內容，並且簡介評估標準。利用實驗的結果來驗證基於移動軌跡模型之異常軌跡偵測方法可行性。並針對異常軌跡偵測方法之正確性進行驗證。最後討論最小支持度及最小軌跡數兩變數對實驗之影響。

Global Positioning System(GPS）almost becomes a powerful tool in our daily life. We can see the application of GPS in all the conveyance. The GPS does not only provide for the positioning and navigation function, but also predict the anomaly trajectory by its trajectory data which the GPS provide.In this paper, we implemented a trajectory model based on anomaly detection by moving objects. After this model built, we could input the real-time trajectory data to the trajectory pattern tree model to see if it shows an anomaly trajectory.In the end of this paper, we described the experimental data and the process with the brief standard of evaluation. We used the result of the experiment to verify the feasibility of detecting the anomaly trajectory, and focused on verifying the precision of detecting the anomaly trajectory. Finally, we discussed how two variables which are the minimum support and the minimum trajectory effecting our experiments.