Eigenbackground Revisited: Can We Model the Background with Eigenvectors?

TL;DR

This paper challenges the traditional Eigenbackground method by demonstrating that using the weakest eigenvectors, rather than the strongest, improves background modeling and reduces artifacts.

Contribution

It introduces a novel approach that utilizes the weakest eigenvectors for background modeling, addressing limitations of the conventional Eigenbackground technique.

Findings

Using weakest eigenvectors reduces artifacts in background modeling.

The proposed method outperforms traditional Eigenbackground in experiments.

Code implementation is available online for reproducibility.

Abstract

Using dominant eigenvectors for background modeling (usually known as Eigenbackground) is a common technique in the literature. However, its results suffer from noticeable artifacts. Thus have been many attempts to reduce the artifacts by making some improvements/enhancement in the Eigenbackground algorithm. In this paper, we show the main problem of the Eigenbackground is in its own core and in fact, it is not a good idea to use strongest eigenvectors for modeling the background. Instead, we propose an alternative solution by exploiting the weakest eigenvectors (which are usually thrown away and treated as garbage data) for background modeling. MATLAB codes are available at \url{https://github.com/mamintoosi/Eigenbackground-Revisited}