Signal Processing, Pattern Recognition and Applications / 779: Computer Graphics and Imaging
Full text: Unavailable
Performing foreground / background segmentation in outdoor scene is very challenging. Starting from a state-of-the-art color-based approach [8], we propose three improvements: the use of the YCoCg c o l o r s p a c e , a spherical association volume, and a cast shadows management approach. U s i n g i m a g e s e q u e n c e s w i t h ground truth at pixel-level, we quantitatively measured the performances of the proposed algorithm and demonstrated that it leads to a reduced processing time with improved detection accuracy. We also introduce a new public dataset of outdoor videos with ground truth.