Automatic Camera Switching in Soccer Game using Decision Tree

Abstract

Cameras management is challenging due in to the need of professional director for controlling and switching between cameras to select the best scéne among them. The aim of this paper is substituting the director of soccer game broadcasting through building automatic camera switching system that works as a TV director that discovers the best scene from a set of cameras and broadcasts it over the network in real time, taking into account the cinematic instructions represented by using the 180-degree rule when filming and avoiding jumping between scenes. Three cameras broadcasting the soccer game in real-time. Every scene in each camera is been analyzed and evaluated by using five parameters are frame direction, previous frame direction, finding ball, number of players, and their position. Many algorithms were used for analyzing the frame such as optical flow for determining the direction of frame, Euclidean distance and Extended Kalman filter for finding the ball, and contour and moment for detecting the players and determining their position. The results of the research were the proposed work was able to create one video stream from three video sources.