Playfield extraction in soccer video based on Lab color space classification

摘要

The growing growth of soccer videos and its enormous popularity around the world has caused the need of creating an automated analysis system. Recognition and extraction of playfield is considered as one of the most basic processes in the analysis of soccer video. Playfield extraction is developed to identify important parts of the shot and remove other additional parts as well as reducing the processing space in order to increase system speed. This paper presents a framework based on Lab color space for classification and extraction playfield in a soccer video. In this method, first, the image is converted to the Lab color space, then the resulting image is classified in order to separate the playfield from other parts. Finally, by using morphological operations, errors and noises are removed and the playfield is extracted as a convex-hull. Experimental results showed that the proposed method is performing very well in the Lab color space, even in the difficult light conditions such as shadows.