Video Moving Object Detection Based on Region Classification Using Three-Order Central Moment

Jin Zheng,Bo Li
DOI: https://doi.org/10.3969/j.issn.1003-9775.2015.05.008
2015-01-01
Journal of Computer-Aided Design & Computer Graphics
Abstract:For the complex outdoor video surveillance, the existing moving object detection methods are dif-ficult to overcome the background interference and accurately detect slow moving objects, as well as small objects simultaneously. These methods have the problem of low accuracy. This paper proposes a moving object detection method based on region classification using three-order central moment. For the different change characteristics of the different pixels in foreground region, cluster region, and background region, which imply the true motion, meaningless motion and background noise, three-order central moment is em-ployed to build the corresponding relationship between the changes of pixel values and the classification of different regions. Therefore, the classifier based on three-order central moments is designed to adaptively classify the scene regions. Accordingly, the moving objects can be detected based on the region classification. The experimental results demonstrate that the proposed method can classify the different regions for outdoor video surveillance effectively. It also can overcome the background interference, e.g., branches swings and water fluctuates, and can detect slow objects and moving small objects accurately.
What problem does this paper attempt to address?