A Robust Moving Object Detection Method for Visual Surveillance Mahfuzul Haque, Manzur Murshed, and Manoranjan Paul Gippsland School of Information Technology, Monash University, Victoria 3842, Australia Email: {Mahfuzul.Haque, Manzur.Murshed, Manoranjan.Paul}@infotech.monash.edu.au Abstract As the importance of human security is increasing day by day, the number of surveillance camera is also growing. But the role of next generation surveillance system is not only capturing video data from different locations and using them for post-incident analysis but also providing support for automatic tracking and early recognition of suspicious activities before any undesirable event happens. This research attempts to propose a robust moving object detection method for automatic visual surveillance . Detection results Conclusion The object detection results and the comparative analysis show that the proposed method performed better than existing methods. Frame 1 Frame 2 Frame t .. K i t i t i t t i t X w X P 1 , , , ) , , ( ) ( ) ( ) ( 2 1 2 / 1 2 / , , 1 | | ) 2 ( 1 ) ( t t T t t X X n t t t e X Gaussian Mixture Model (GMM) for each pixel Input frames First Frame Test Frame Ideal Result Actual Result Comparative analysis of the experimental results The technique behind the scene Moving Object Detection Toolkit The image shown in the header has been taken from http://www.informationliberation.com The test sequences have been taken from the PETS database (http://www.cvg.rdg.ac.uk/slides/pets.html) Tool for analysing underlying statistical model Note: Fewer false classifications refer preferred detection method. A pixel model is constructed and updated for each pixel which maintains a Mixture of Gaussian distributions to address the multi-modal distribution caused by illumination variations or a repetitive background motions due to moving leaves on a tree or moving clouds. 0 50 100 150 200 250 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Test Sequence False Classification (thousands) GMM (Stauffer) GMM (Lee) Proposed Method