Open Access iconOpen Access



Background Subtraction in Surveillance Systems Using Local Spectral Histograms and Linear Regression

S. Hariharan1,*, R. Venkatesan2

1 Coimbatore Institute of Technology, Department of Computer Science and Engineering, Coimbatore 641 014, India
2 PSG College of Technology, Department of Computer Science and Engineering, Coimbatore 641 004, India

* Corresponding Author: S. Hariharan. Email: email

Intelligent Automation & Soft Computing 2022, 34(1), 407-422.


Background subtraction is a fundamental and crucial task for computer vision-based automatic video analysis due to various challenging situations that occur in real-world scenarios. This paper presents a novel background subtraction method by estimating the background model using linear regression and local spectral histogram which captures combined spectral and texture features. Different linear filters are applied on the image window centered at each pixel location and the features are captured via these filter responses. Each feature has been approximated by a linear combination of two representative features, each of which corresponds to either a background or a foreground pixel. These representative features have been identified using K-means clustering, which is used in background modeling using a least square method. Constraints have been introduced in the least square solution to make it robust in noisy environments. Experiments on existing datasets show that our proposed method outperforms the methods in the literature, with an overall accuracy of 92%.


Cite This Article

S. Hariharan and R. Venkatesan, "Background subtraction in surveillance systems using local spectral histograms and linear regression," Intelligent Automation & Soft Computing, vol. 34, no.1, pp. 407–422, 2022.

cc This work is licensed under a Creative Commons Attribution 4.0 International License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
  • 1000


  • 847


  • 0


Share Link