In football, various situations arise in which referees have difficulty making correct judgements without additional assistance. To that end, various technologies are being employed with increasing frequency.
The thesis presents the implementation of a design of a multi-camera goal detection system using online cameras. We set up a system in controlled laboratory circumstances using cameras which are calibrated in a single coordinate system. Then the input images are segmented using background modelling and we search for the centre of the object, in this case a yellow ball. The positions of the ball and the goal line are determined using triangulation in a global coordinate system, which is then converted to 3D space, making it possible to determine whether the ball has crossed the line. We can also assess the accuracy of the operation of the system and any errors that occur. The final average error in detecting goals is less than 1.5 mm, which is encouraging for a system that uses consumer-level cameras.
|
