Moving Cast Shadow Detection
We developed a novel algorithm to detect and remove moving cast shadows in a video sequence.
Background & Objective: Cast shadows poses one of the most challenging problems in many vision tasks, especially in object tracking, by distorting the true shape and color properties of the target objects. They correspond to the areas in the background scene that are blocked from the light source. It is essential to eliminate only cast shadows since removal of self shadows, which are the parts of the object that are not illuminated, will result in incomplete object silhouettes.
Technical Discussion: We remove cast shadows by taking advantage of the statistical prevalence of the shadowed regions over the object regions. We model shadows using multivariate Gaussians. We apply a weak classifier as a pre-filter. We project shadow models into a quantized color space to update a shadow flow function. We use shadow flow, background models, and current frame to determine the shadow and object regions.
This method has several advantages: 1) It does not require a color space transformation. We pose the problem in the RGB color space, and we can carry out the same analysis in other Cartesian color spaces as well. 2) Our data driven method dynamically adapts the shadow models to the changing shadow conditions. In other words, accuracy of our method is not limited by the preset threshold values, which is a major drawback of the existing approaches. The accuracy of this method improves as it process more video frames. 3) Furthermore, it does not assume any 3D models for the target objects or tracking of the cast shadows between frames.
Publications:
Technology Area: Computer Vision
Modification Date: June 16, 2006