A High Accuracy Camera Calibration Method for Sport Videos

Camera calibration for sport videos enables precise and natural delivery of graphics on video footage and several other special effects. This in turns substantially improves the visual experience in the audience and facilitates sports analysis within or after the live show. In this paper, we propose a high accuracy camera calibration method for sport videos. First, we generate a homography database by uniformly sampling camera parameters. This database includes more than 91 thousand different homography matrices. Then, we use the conditional generative adversarial network (cGAN) to achieve semantic segmentation splitting the broadcast frames into four classes. In a subsequent processing step, we build an effective feature extraction network to extract the feature of semantic segmented images. After that, we search for the feature in the database to find the best matching homography. Finally, we refine the homography by image alignment. In a comprehensive evaluation using the 2014 World Cup dataset, our method outperforms other state-of-the-art techniques.

I. Introduction

Camera calibration is a fundamental task in computer vision. It can be applied to solve many other problems including natural graphics, tracking of players and augmented reality (AR) enhanced broadcast [1], [2]. With the advent of deep-learning technologies sport graphics has experienced a healthy revival over the last few years [3]–[6]. Here the aim is to improve the accuracy of camera calibration for improved video delivery. Indeed, accurate camera calibration for sport and other broadcast videos can not only give audiences a better visual experience but also facilitate sports analysis within or after the live show. In addition, camera calibration enables professional coaches to exploit statistics for enhanced tactics in the field. Companies like ESPN and Stats Perform, base their intelligent broadcast on camera calibration. In this paper, we introduce a high accuracy camera calibration method for sport videos. Although we focus on soccer videos, the proposed technique can be adapted to other types of “pitch sports”.