Given the location of shadows, how can we obtain highquality shadow-free images? Several methods have been proposed so far, but they either introduce artifacts or can be difficult to implement. We propose here a simple method that results in virtually error and shadow-free images in a very short time. Our approach is based on the insight that shadow regions differ from their shadow-free counterparts by a single scaling factor. We derive a robust method to obtain that factor. We show that for complex scenes - containing many disjointed shadow regions- our new method is faster and more robust than others previously published. The method delivers good performance on a variety of outdoor images.
|Number of pages||4|
|Publication status||Published - Aug 2006|
|Event||18th International Conference on Pattern Recognition - Hong Kong|
Duration: 20 Aug 2006 → 24 Aug 2006
|Conference||18th International Conference on Pattern Recognition|
|Period||20/08/06 → 24/08/06|