Abstract
Background modelling is a method of detecting moving objects from the difference between current frame and a reference frame. Benefits of background modelling and object detection are uncountable, and have enormous applications in the area of artificial intelligence, video survillance, medicine and some of security related applications. Popular background modelling methods are Gaussian Mixture Model(GMM), Adaptive Gaussian Mixture Model (AGMM), Kernel Density Estimator (KDE), Codebook, Visual Background Extractor(VIBE), Self-Organizing Background Subtraction (SOBS), all these models suffers from illumination changes and dynamic background. In this paper, GMM and SOBS algorithms are implemented using MATLAB software for object detection and tracking, and evaluated results using performance parameters such as recall, specificity, False positive Rate (FPR), False Negative Rate (FNR), precision, Percentage of Wrong Classification (PWC), F-measure. Obtained results shows that 9.71% improvement in specificity of GMM and 2.97% improvement in specificity of SOBS.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have