The analysis and exploration of auroral dynamics are very significant for studying the auroral mechanism. In this paper, a method based on an auroral dynamic process for automatically detecting an auroral event is proposed. We first obtained the local motion information using a multi-scale fluid flow estimator. Then the auroral sequence was represented by the spatiotemporal statistics of a local motion vector. Finally, auroral event-based detection was automatically achieved. The experimental results showed our methods could effectively and accurately detect the required auroral events, and were not dependent on any specified auroral event. The proposed method makes it feasible to statistically analyze a large number of continuous observations based on the auroral dynamic process.