Low-Rank Representation with Contextual Regularization for Moving Object Detection in Big Surveillance Video Data

Modern video surveillance benefits greatly from advanced wireless imaging sensors and cloud data storage; thus, a vast amount of data is generated every second. Surveillance videos have thus become one of the biggest sources of unstructured data. Because a vast amount of surveillance videos is continuously and quickly produced at multiple locations, moving object detection in such a vast amount of these videos by using traditional detection methods is a challenging task. This paper presents a novel model that detects moving objects from such data sets based on low-rank representation with contextual regularization. Quantitative and qualitative assessments indicated that the proposed model significantly outperformed existing state-of-the-art moving object detection methods.