A fog based ball tracking (FB2T) system using intelligent ball bees

Abstract

Soccer is the most popular group game; it has a large number of fans all over the world. Although each complete soccer match lasts around one hour and a half, only have very few scenes attracting audiences been perfectly filmed. This filming has usually been done using static (fixed) cameras, and they cannot provide the same level of accuracy or entertainment given by mobile devices. Taking into our consideration that there are no schemes using mobile devices have been proposed to solve this problem till now. As periodic communication generates a huge amount of data; typical storage, computation, and communication resources are required. Hence, this paper aims at introducing a Fog computing mechanism that takes into consideration the requitrements of special mobility, low latency and location awareness. Our solution based on controlling the movement of mobile cameras mounted on a fleet of mobile bees. A bee is an autonomous camera-drone, which gives the audience the feeling of being a part of such sports competition. It is a special type of Unmanned Aerial Vehicle (UAV) equipped with a built-in processing unit, memory, high-speed camera and transceiver. The bee has the ability to film the movement of the ball, which has to be followed and filmed via object tracking principle. In fact, detecting and tracking the ball from the broadcast soccer video constitute a major challenge. In soccer matches, the ball moves most of the time and it is frequently occluded while its size and shape appearance vary over the time and between cameras. Moreover, the feature-based tracking methods are used to judge whether or not a sole object is the target as the features of the ball might be changed fast over frames and then we cannot manage to distinguish the ball from other objects by using these methods. Thus, the current study demonstrates an innovative technique for tracking a soccer ball from mobile cameras fixed on multiple ball bees.