Cast shadows from moving objects reduce the general ability of robust classification and tracking of these objects, in outdoor surveillance applications. A method for segmentation of cast shadows is proposed, combining statistical features with a new similarity feature, derived from a physics-based model. The new method is compared to a reference method, and found to improve performance significantly, based on a test set of real-world examples.
|Title of host publication||Den 14. Danske Konference i Mønstergenkendelse og Billedanalyse|
|Publisher||Informatics and Mathematical Modelling, Technical University of Denmark, DTU|
|Publication status||Published - 2005|