Pfinder: Real-Time Tracking of the Human Body
IEEE Transactions on Pattern Analysis and Machine Intelligence
Real-time adaptive background segmentation
ICME '03 Proceedings of the 2003 International Conference on Multimedia and Expo - Volume 3 (ICME '03) - Volume 03
Robust background subtraction with foreground validation for urban traffic video
EURASIP Journal on Applied Signal Processing
A Background Reconstruction Algorithm Based on Two-Threshold Sequential Clustering
CCCM '08 Proceedings of the 2008 ISECS International Colloquium on Computing, Communication, Control, and Management - Volume 01
Journal on Image and Video Processing - Special issue on advanced video-based surveillance
Fusing color and texture features for background model
FSKD'06 Proceedings of the Third international conference on Fuzzy Systems and Knowledge Discovery
| Hi-index | 0.00 |
In video surveillance, the detected foreground is deformed or shapeless in cluttered or dynamic backgrounds. For this propose, we present an improvement method that can work in real time for silhouette determination of moving objects and dynamic background construction in such conditions. At first, image sequences are analysed pixel by pixel using improved basic sequential clustering algorithm. Then, clusters having high weight are used to estimate the background and pixels not belonging to the background clusters are labeled as foreground. Finally, the extracted foreground is treated from possible noises in a Markov Random Field framework. The original method is improved from the ghost effect problem, which drops some regions from the esteemed background and appears them as detected foreground, by adding clusters keeping the past stat of those regions from deviation to foreground detection. In addition, space memory is optimised by deleting old clusters that not updating after a timeout. The experiments show better results than the classical method in cluttered and multi background circumstances.