A method for game analysis based on dominant region
In any football game, space management by players is one of the most important factors to exercise control over the game. In game analysis, the amount of player space and its location, in addition to its changing characteristics throughout the game are essential information. However, there are few methods to obtain such data regarding player space quantitatively and visually. Therefore, an automated system for analysing and quantifying this space from video sequences has been identified as necessary. In this paper, we describe novel methods for extracting quantitative information about this space from video sequences and for analysing a game utilising the extracted results. Those methods are based on the concept of ‘dominant region’ (Taki and Hasegawa, 2000) and a new measure called ‘time occupancy rate’. Also, the results of an analysis of an actual game using these methods are presented, and the effectiveness of our system is demonstrated. 2. DOMINANT REGION The immediate region around any given player can be considered a kind of sphere of potential influence of that player. Generally, given a set of points in a space, the spatial territory of each point can be expressed by the Voronoi region (Okabe et al., 1992). In actual games, the player’s sphere of influence is changed according to the direction and speed of their movements. This change can be observed through frame by frame video analysis. Therefore, the sphere of influence is formulated by replacing the distance function of the Voronoi region with a time function. We call this the ‘dominant region’. The dominant region of a player is defined as a region where the player can arrive earlier than any other players, on condition that the player moves at their maximal speed. Also, the physical ability of each player can be reflected by the dominant regions. Figure 1 shows an example of the changes of dominant regions in a case where a player receives the ball behind an opponent player with a checking run. In the Figures 1(a) to 1(f), the background shows the dominant region of each player, and each dominant region is distinguished by different colours. We can easily observe that player A will be
able to receive the ball, because the space behind player B is becoming the dominant region of player A.