8.1 Introduction

Sports scenarios present an interesting challenge for visual surveillance applications. Here, we describe a system, and a set of techniques, for tracking players in a football (soccer) environment. The system input is video data from static cameras with overlapping fields-of-view at a football stadium. The output is the real-world, real-time positions of football players during a match. In this chapter, we discuss the problems and solutions that have arisen whilst designing systems and algorithms for this purpose.

The overall application output is the positions of players, and the ball, during a football match. This output can be used live, for entertainment augmenting digital TV or low-bandwidth match play animations for Web or wireless display and played back for analysis of fitness and tactics of the teams and players. At this stage, the technology is not sufficient to allow reliable recognition of the identities of each player in the game, so at present we aim only at automatic recognition of the team of each player. Indeed, there are five different uniforms of clothing on the pitch (two teams, two goalkeepers and the referees), so we aim to recognise these different categories of player.

Several research projects on tracking soccer players have published results in this field. Intille and Bobick [ ^[1]] track players, using the concept of a closed-world, in the broadcast TV footage of American football games. This concept is defined as space time region of an image sequence in which the complete taxonomy...