Shadow Classification and Evaluation for Soccer Player Detection

Abstract

In a football stadium environment with multiple overhead floodlights, many protruding shadows can be observed originating from each of the targets. To successfully track individual targets, it is essential to achieve an accurate representation of the foreground. Unfortunately, many of the existing techniques are sensitive to shadows, falsely classifying them as foreground. In this work an unsupervised learning procedure that determines the RGB colour distributions of the foreground and shadow classes of feature data is proposed. A novel skelatonisation and spatial filtering process is developed for identifying components in the foreground segmentation that are most-likely to belong to each class of feature. A pixel classification mechanism is obtained at by approximating both classes of feature data by N Gaussian parametric models. To assess our technique's performance and reliability, a comparison is made with other published works.

A vision-based system to support tactical and physical analyses in futsal

“…The detection step using color-based techniques is usually fast, but it is very sensitive to illumination variations, which may reduce their robustness and precision. Moreover, the techniques in [38,31] demand on a laborious training step.…”

“…Some works are based on fixed cameras [30,10,11,24,18,20,32,17,46,38], since they can capture, in most cases, all players actions in the game region, while moving cameras or broadcast images, on the other hand, can not always view all players in the scene during the entire game, which causes the loss of certain actions. Nevertheless, many works in that field use these types of image sources [22,35,49,15,23,31], as they are most likely easier to obtain.…”

“…In [13,22,17,15,46], a model that represent the game region is built based on color information, that is, the predominant game region color is used to create a color model of the background, allowing the extraction of regions that contain players. Similarly, the authors in [38,31] use histograms and color distributions to build a model that represents the player and it is used on the segmentation process. This model is built from players samples manually collected in a training phase.…”

The Perception and Content of Cast Shadows: An Interdisciplinary Review

“…Given this rough shadow segmentation they then build a Gaussian model of the color distribution of the shadow pixels, allowing color based refinement of the shadow model. Renno, Orwell, Thirde, and Jones (2004) describe a shadow detection technique which uses strong spatial information to augment a color based shadow segmentation. In this article they deal with the characteristic quadruple shadows cast by football players under floodlights, and a novel skeletonization approach is used to distinguish those foreground detections due to the cast shadows (which appear on the floor) and those which are due to actual foreground motion.…”

Players and Ball Detection in Soccer Videos Based on Color Segmentation and Shape Analysis

“…Recently, skeleton analysis has been employed for object recognition [5,12], shadow removal [13] and shape filtering [2,14]. This paper proposes a shape analysis-based approach to identify the players and the ball from the roughly extracted foreground, which is obtained by a trained color histogram-based playfield detector and connected component analysis.…”