A Robust Algorithm for Shadow Removal of Foreground Detection in Video Surveillance

We proposed an accurate algorithm to prevent moving shadows from being misclassified as part of moving objects in video target segmentation in this paper. Firstly, moving objects were achieved through background subtraction using adaptive Gaussian mixture models. Then, moving shadows were eliminated by a shadow detection algorithm. Finally, we performed a morphological reconstruction algorithm to recover the foreground distorted after shadow removal process. The experimental results proved its validity and accuracy in various fixed outdoor video scenes.