Recognition of sport players’ numbers using fast color segmentation C´ edric Verleysen * and Christophe De Vleeschouwer ICTEAM institute, Universit´ e catholique de Louvain, Louvain-la-Neuve, Belgium ABSTRACT This paper builds on a prior work for player detection, and proposes an efficient and effective method to dis- tinguish among players based on the numbers printed on their jerseys. To extract the numbers, the dominant colors of the jersey are learnt during an initial phase and used to speed up the segmentation of the candidate digit regions. An additional set of criteria, considering the relative position and size (compared to the player bounding box) and the density (compared to the digit rectangular support) of the digit, are used to filter out the regions that obviously do not correspond to a digit. Once the plausible digit regions have been extracted, their recognition is based on feature-based classification. A number of original features are proposed to increase the robustness against digit appearance changes, resulting from the font thickness variability and from the de- formations of the jersey during the game. Finally, the efficiency and the effectiveness of the proposed method are demonstrated on a real-life basketball dataset. It shows that the proposed segmentation runs about ten times faster than the mean-shift algorithm, but also outlines that the proposed additional features significantly increase the digit recognition accuracy. Despite significant deformations, 40% of the samples, that can be visually recognized as digits, are well classified as numbers. Out of these classified samples, more than 80% of them are correctly recognized. Besides, more than 95% of the samples, that are not numbers, are correctly identified as non-numbers. Keywords: Color segmentation, k-means, digit recognition, feature-based classification, SVM. 1. INTRODUCTION In recent years, digit and character recognition have become very popular fields in image processing. Among the principal applications, one could think about car plate recognition 1 , automatic sorting of post letters 2 , digital archiving of books 3 , etc. Automatic character recognition (OCR) is also a powerful tool that gives to the computers the ability to autonomously recognize objects that have been labelled in a structured manner. Such interpretation is for example required for the autonomous and low-cost production of personalized summaries of sport events 4, 5 . In such context, the recognition of the digits printed on the players’ jerseys appears to complete conventionnal people detection and tracking algorithms to understand the scene. In particular, our paper builds on earlier works 6, 7 to detect the players, and proposes an efficient approach to identify them based on the segmentation and recognition of their jerseys’ number. Figure 1. Digit recognition is divided into two stages: the digit extraction and its classification. * cedric.verleysen@uclouvain.be