Open AccessAn Improvement for Background Modelling using a Mixture of Gaussian and Region Growing in Moving Objects Detection
Author(s) -
Moch Arief Soeleman,
Sandhya Yogi,
Aris Nurhindarto,
Muslih,
Muljono Muljono,
W Karis,
Ricardus Anggi Pramunendar
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1430/1/012032
Subject(s) - object detection , computer science , artificial intelligence , gaussian , object (grammar) , statistical model , background subtraction , position (finance) , foreground detection , measure (data warehouse) , computer vision , gaussian network model , mixture model , statistical power , pattern recognition (psychology) , data mining , mathematics , statistics , pixel , physics , finance , quantum mechanics , economics
Most of the research on object detection that uses discussing background is only necessary for the decision of the background model by assuming all statistical objects are part of the background. This caused a debate called “foreground aperture,” where statistical purposes that represent moving backgrounds, the position of the initial object will be detected as false detection. Related to the new background modeling is needed to overcome this problem. This paper proposes how to obtain an efficient background model that meets real requirements. As a background model, the Mixed-of-Gaussian Model (MOG) was adopted and refined with some improvements with the Region Growing method. Evaluation to measure the performance of the MOG algorithm and Region Growing by using a detection rate to calculate the number of right and wrong in the detection of moving objects.