Foreground object detection for video surveillance by fuzzy logic based estimation of pixel illumination states. (11th September 2018)

• Record Type: Journal Article
• Authors: Molina-Cabello, Miguel A; López-Rubio, Ezequiel; M Luque-Baena, Rafael; Domínguez, Enrique; Palomo, Esteban J
• Is Part Of: Logic journal of the IGPL. Volume 26:Number 6(2018:Dec.)
• Page Start: 593
• Page End: 604
• Publication Date: 2018-09-11
• Subjects: Foreground detection -- probabilistic self-organising maps -- background features -- fuzzy rule-based system; Logic, Symbolic and mathematical -- Periodicals; 511.3
• DOI: 10.1093/jigpal/jzy024
• Languages: English
• ISSNs: 1367-0751

Abstract

Abstract: In this paper, a probabilistic self-organising map-based model combined with a fuzzy logic model is proposed for foreground object detection in video sequences. Our method consists of two different parts. First, a baseline model based on probabilistic self-organising maps is applied, where a set of a suitable set of characteristic pixel features is chosen. In this stage, a uniform distribution to represent the foreground is used. Then, the output of the baseline model is supplied to a fuzzy logic subsystem, which is able to estimate the illumination states of the pixels. Experimental results demonstrate the good performance of our proposal compared to some competing methods.