CUI Lahore Repository

Foreground Extraction by using Kernel Density Estimation

Show simple item record

dc.contributor.author Ahmed, Hafiz Syed Bilal
dc.contributor.author Danish, Shoaib
dc.contributor.author Ali, Zulfiqar
dc.date.accessioned 2018-10-05T07:44:20Z
dc.date.available 2018-10-05T07:44:20Z
dc.date.issued 2017-05-16
dc.identifier.uri http://hdl.handle.net/123456789/793
dc.description.abstract There have been a vast amount of studies on background modelling to detect moving objects. Recent reviews showed that Kernel Density Estimation (KDE) and Gaussian Mixture Model (GMM) perform about equally best among possible background modelling. In this project, we will work on the background modeling to detect the moving objects and separate the background and foreground objects from a particular video. In KDE, the selection of different kernel functions and their bandwidth is important in that they determine the underlying probability distribution and thus the quality of background modeling. In this project, we will pick four best kernel functions and evaluate their results by comparing their pictorial and graphical results. Then we will test these kernels on an artificial video and check its results and see which one is best among them. Till now, we have successfully evaluated the results of four best kernel functions which are Epanechnikov, Bi-weight, Triangular and Tricube and compared their results. We have also developed a new kernel function named Ultra-fore and compared this function with the best four functions mentioned above. The results we have got are very satisfactory. All tests were done on real videos with varying background dynamics and results were analysed both quantitatively and qualitatively. In this report, we will discuss about Epanechnikov, Bi-weight, Triangular and Tricube kernel functions and a new kernel function named Ultra-fore and its results. We will also discuss about their bandwidths and threshold values. en_US
dc.language.iso en en_US
dc.publisher COMSATS University Islamabad, Lahore Campus en_US
dc.subject Computer Science en_US
dc.subject Triangular kernel function, Tricube kernel function, Bi-weight kernel function, Epanechnikov kernel function, Ultra-fore kernel function. en_US
dc.title Foreground Extraction by using Kernel Density Estimation en_US
dc.type Learning Object en_US


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace


Browse

My Account