Shadow Removal Using B-Spline for Edges Reconstruction in Surveillance
Redwan A.K. Noaman, Mohd Anuar Md Ali, Kasmiran Jumari · Procedia Engineering · 2012
Shadows are one of the major problems in objects detection and tracking for automated video surveillance systems. In such systems accurate detection is required even under variable weather conditions. The majority of shadow removal algorithms in the literature suffer from misclassification errors because of inaccurate shadow detection, thus resulting in distorted object shapes. In this paper, a novel technique is proposed for shadow detection and removal using both edge detecting and B-spline algorithm. This technique is capable of correcting such errors by using B-spline algorithm for object's edges reconstruction so that the object shapes remain well defined along most part of their contours. Experimental results in PETS2006 benchmark dataset show that the proposed technique increases the shadow detection reliability compared to some existing shadow detection algorithms.