Home // International Journal On Advances in Systems and Measurements, volume 10, numbers 1 and 2, 2017 // View article

Comparative Evaluation of Background Subtraction Algorithms for High Performance Embedded Systems

Authors:
Lorena Guachi
Giuseppe Cocorullo
Pasquale Corsonello
Fabio Frustaci
Stefania Perri

Keywords: Real-Time; Image processing; Background subtraction; Segmentation

Abstract:
Background Subtraction technique is widely used in surveillance systems to identify moving objects. Although color features have been extensively used in several background subtraction algorithms, demonstrating high efficiency and performances, in actual real-time applications the background subtraction performance is still a challenge due to high computational requirements. In this paper, two approaches and their optimized versions are evaluated to implement high-performance background subtraction algorithms for real-time applications. Gaussian Mixture Model and the Multimodal Background Subtraction are characterized by two different color descriptors: Gray scale and H color invariant combined with Gray scale information respectively. Different experimental analysis allows evaluating the efficiency in terms of computational complexity and accuracy for outdoor and indoor environments. Experimental tests demonstrated that the Multimodal Background Subtraction approach with its variants is established as affordable for real-time applications and particularly suitable on hardware platforms with on-board memory and limited computational resources.

Pages: 35 to 44

Copyright: Copyright (c) to authors, 2017. Used with permission.

Publication date: June 30, 2017

Published in: journal

ISSN: 1942-261x