Authors : Tawfiq Abdulkhaleq Al Asadi and Fanar Ali Joda
Abstract: Object detection and moving cast shadow removal are crucial steps in video surveillance system. In this study, we have proposed background modelling algorithm for objects detection has ability to update threshold when illumination is changed which leads to increase the performance of surveillance system instead of other methods that suffer from a general threshold for all cases which fail when there is a change in illumination. In addition, moving cast shadow removal algorithm is proposed it consists of two stages, the first one, candidate shadow regions are obtained by using weak detector. The second one, correlation is calculated between reference and current frames of candidate shadow regions, this stage aims to classify pixels of candidate shadow regions into shadow and foreground object.
Tawfiq Abdulkhaleq Al Asadi and Fanar Ali Joda, 2017. Adaptive Background Modelling and Shadow Removal for Video Surveillance System. Journal of Engineering and Applied Sciences, 12: 7689-7695.