Background Modeling Using Adaptive Pixelwise Kernel Variances in a Hybrid Feature Space

TL;DR

This paper introduces an adaptive pixelwise kernel variance method in a hybrid feature space for background modeling, combining probabilistic models and complex features to improve background subtraction accuracy.

Contribution

It proposes a heuristic for adaptive kernel variances at each pixel, enhancing background modeling efficiency and accuracy over previous methods.

Findings

Significant improvement on a standard backgrounding benchmark.

Adaptive kernel variances outperform fixed variance models.

Combining complex features with adaptive kernels yields better results.

Abstract

Recent work on background subtraction has shown developments on two major fronts. In one, there has been increasing sophistication of probabilistic models, from mixtures of Gaussians at each pixel [7], to kernel density estimates at each pixel [1], and more recently to joint domainrange density estimates that incorporate spatial information [6]. Another line of work has shown the benefits of increasingly complex feature representations, including the use of texture information, local binary patterns, and recently scale-invariant local ternary patterns [4]. In this work, we use joint domain-range based estimates for background and foreground scores and show that dynamically choosing kernel variances in our kernel estimates at each individual pixel can significantly improve results. We give a heuristic method for selectively applying the adaptive kernel calculations which is nearly as accurate as the full procedure but runs much faster. We combine these modeling improvements with recently developed complex features [4] and show significant improvements on a standard backgrounding benchmark.