GLABM: Gamma Corrected Layered Adaptive Background Model for Outdoor Scenes
Kousuke Sakamoto, Hiroki Yoshimura, Masashi Nishiyama, Yoshio Iwai · 2018
This paper proposes a method for pixel-based background subtraction with adaptive gamma correction and a layered adaptive background models (LABM). Background oscillation and shadow removal are challenging problems for background subtraction. To tackle these problems, we extend the adaptive background model (ABM) to the multi-layered models. ABM has the disadvantage of assuming a linear RGB color space, so we implement the gamma estimation as a preprocessing step for LABM. We demonstrate the performance of the proposed method by comparing with other pixel-based background subtraction methods.