EVALUATION OF SHADOW CLASSIFICATION TECHNlQUES FOR OBJECT DETECTION AND TRACKING
J. R. Renno, James Orwell, Grueme A. Jones · 2004
In a football stadium environment with multiple overhead lloodlights. many protruding shadows can he observcd originating from each of the targets. To successfully track individual targets, it is essential to achieve an accurate representation of the foreground. Many of the cxisting techniques are sensitive 10 shadows. falsely classifying shadows as foreground. This work presents four different techniques associated with shadow classification. Three of the classifier’s originate from the review material whilst the fourth is a novel application of a real-time implementation of the k-nearest neighburrr algorithm to shadow identification. To assess the performance for each of the classifiers four quantitative evaluation metrics are proposed. Using each of the evaluation metrics, we will discuss the performance of each classifier’s segmentation results as well as assess their impact on the tracking performances.