Pass maps that visualize passes in a soccer match are the fundamental tools to analyze attacks by the opponent team. However, many passes are attempted in a match, so drawing the trajectories of all passes causes heavy cluttering in the visualization result. One of the filtering strategies is to form clusters by positions, passer, and receiver of passes, but a simple clustering algorithm sometimes fails to form reasonable results. This paper proposes an enhanced algorithm based on the adaptive DBSCAN clustering. It adaptively increases the maximum distance between two passes so that our method can reduce the number of passes not included in the cluster. Experimental results show that the method can make more clusters, increasing the number of passes in clusters.
|