Title
Tracking Futsal Players With A Wide-Angle Lens Camera: Accuracy Analysis Of The Radial Distortion Correction Based On An Improved Hough Transform Algorithm
Abstract
The aim of this study was to assess the applicability and reliability of a single wide-angle lens GoPro r camera for tracking and kinematics analysis of futsal players. An official game of a Brazilian professional team in the quarter-final round of the 2013 Sao Paulo futsal league was recorded by four digital video cameras (30 Hz; 720 x 480) placed at the highest points of the court (40 x 20 m; FIFA standard). We compared tracking performed with three cameras equipped with conventional lenses to a GoPro r camera with a wide-angle lens with a 170 degrees field of view. Manual and automatic tracking of players trajectories (n = 5; 23.20 +/- 2.39 years) during competition and in a controlled environment were performed. We found that the root-mean-square error calculated to determine the position on the court based on images captured with a GoPro r camera was 0.32 m and for the velocity was 0.71 m s-1. A 5-min excerpt of the game showed an average difference of 2.11% between the total distance covered obtained by tracking with the GoPro r with radial distortion correction (438.97 +/- 164.65 m), and cameras with conventional lenses (449.54 +/- 170.92 m). Temporal correlation analysis confirmed the great similarity between the reconstructed trajectories obtained by these cameras (r = 0.99 for movements on both x and y axes). We showed that a single GoPro r Hero3+ can provide reliable kinematic analysis of computational tracking based on videogrammetry for futsal, using the Hough transform and division model technique for radial distortion correction.
Year
DOI
Venue
2017
10.1080/21681163.2015.1072055
COMPUTER METHODS IN BIOMECHANICS AND BIOMEDICAL ENGINEERING-IMAGING AND VISUALIZATION
Keywords
Field
DocType
Radial distortion, Hough transform, GoPro, automatic tracking, futsal, biomechanics
Distortion (optics),Field of view,Computer vision,Digital video,Kinematics,Wide-angle lens,Computer graphics (images),Computer science,Hough transform,Lens (optics),Artificial intelligence
Journal
Volume
Issue
ISSN
5
3
2168-1163
Citations 
PageRank 
References 
1
0.36
6
Authors
9