Introduction - If you have any usage issues, please Google them yourself
Robustnesstochangesinilluminationcon ditionsaswellas viewing perspectives is an im portant requirement formany a computer vision pplications. One of the key fac-ors in enhancin g the robustness of dynamic scene analy-sis is not hat of accurate and reliable means for de shadow- ection. Shadowdetectioniscriticalforcorr ectobjectde- ection in image sequences. Many a lgorithms have beenproposed in the literature that deal with shadows. How-ever, acomparativeevaluationoftheexistingappro achesisstill lacking. In this paper, the full range of problems un-derlyingtheshad owdetectionareidenti edanddiscussed.Wecla ssifytheproposedsolutionstothisproblemus ingaaxonomyoffourmainclasses. calleddeterministicmodeland non-model base d and statistical parametric and non-parametr id. Novelquantitative (det