Recently the video surveillance market has developed rapidly, but judging whether there is abnormal behavior in the video relying on manpower is too expensive. Therefore, a method is needed to identify the abnormal behavior automatically. Scholars at home and abroad have done in-depth research on video abnormal events detection in different scenarios. However, the current detection technology still needs improvement in the speed of the algorithm. From this point of view, this paper proposes a video abnormal event detection method based on hierarchical clustering.

In order to construct the sparse coefficient matrix more accurately and quickly, the hierarchical clustering is introduced into sparse coding in this paper. And the structure information of the sparse coefficient matrix is used as the clustering criteria, which improves the standard group sparse coding method. In addition, the BK-SVD algorithm is used to train the dictionary so that we can further improve the speed of the algorithm through dictionary division. In the experimental part, we prove that the proposed algorithm has great performance in frame level and pixel level in MATLAB environment.