Ferramentas para análise de partidas de futsal utilizando processamento de imagens e visão computacional

Abstract

The use of computer systems in sports has had a significant growth in last decade. Tools that help reduce errors such as computer vision to identify whether the ball has bounced in or out of a tennis court or whether or not it has passed the goal line in football field are widely known. But, although less popular, there are also systems that help each athlete or team quantify their performance, such as distances traveled, speeds attained, and positions where each athlete was on the court or field. However, futsal, a very popular sport in Brazil, still does not receive much attention regarding the development of these systems. In this work, tools will be developed and applications will be presented about image processing and computer vision for futsal game analysis. The approach was illustrated through video recordings of matches of the Campeonato Paranaense de Futsal (Serie Ouro), from which were automatically extracted the movements of the athletes of the team Pato Futsal from Pato Branco - PR, we decided to ignore the goalkeeper. Three video cut were analysed in this study and, from them, data visualization were generated to help coaches in their physical and tatical analysis. The videos, identified as I, II and III, showed different visualizations. Video I presented the best results, where all the players were identified and tracked. Video II showed tracking for three players and video III for two players.