Advanced motion detection for intelligent video surveillance systems
Fan‐Chieh Cheng, Shih-Chia Huang, Shanq-Jang Ruan · 2010
In this paper, we propose a novel background subtraction method that makes use of spectral, spatial, and temporal features extracted from the video sequence in determination of the best background candidates for background modeling. As the final step of our process, the binary moving object detection mask is computed using prompt background subtraction with our proposed background model. The overall results of these analyses thus demonstrate that our proposed method substantially outperforms existing methods by an F1 metric accuracy rate increase of up to 79%.