COLOR IMAGES SEGMENTATION ALGORITHMS DURING A SPORTS MEETING: APPLICATION TO SOCCER VIDEO IMAGES
Mourad Moussa Jlassi, Ali Douik, Hassani Messaoud · Journal of Circuits Systems and Computers · 2010
The choice of the best color system representation is an efficient technique in background modeling and a fundamental task in detecting foreground objects on both static and dynamic scenes. In this paper, we present some algorithms consecrated for images color segmentation's issued from a soccer video. This manuscript is prominent for many next computers vision studies. Indeed, in this paper many segmentation techniques using pixel values are developed (statistical, fuzzy tools) and gives a high robustness level even in presence of significant shadows and highlights. The use of RGB color space to represent a pixel is not obvious to lead to valuable results in certain situations, for this reason, we proposed a new approach using color to improve players detection in images by determining the three significant color components on the one hand, and using statistical theory to model the background on the other hand. In fact the use of color is widely employed by many researchers. Also this paper presents experimental results of soccer scenes, which will evaluate the proposed methods in order to extract players and eliminate the effects of shadows, and to keep the best levels.