A Brief Review on background substraction

EasyChair Preprint 5950

Abstract

Background subtraction is the process of separating the foreground from background of an image. The applications of background subtraction range widely in important areas, viz- intelligent video surveillance, intelligent visual observation of animals and insects, optical motion capture, human machine interaction etc. The techniques available for background subtraction are broadly grouped under the categories of traditional background modelling or recent background modelling, pixel or regional level etc. Further, the categorization is also done based on specific challenges to be addressed, which include- statistical model, cluster model, neural network model etc. The issues with background subtraction are diverse-noisy image, camera jitter, illumination changes etc. In this paper, we discuss briefly on the various stages of background subtraction, its challenges and techniques apt in handling those challenges. Also, we look at various datasets under the category of traditional and recent dataset which poses multi-pronged challenges.

Keyphrases: Background maintenance, Background modelling, background initialization, background subtraction, foreground detection

@booklet{EasyChair:5950,
  author    = {B. G. Vishruth and M. T. Gopalakrishna and J. Megha},
  title     = {A Brief Review on background substraction},
  howpublished = {EasyChair Preprint 5950},
  year      = {EasyChair, 2021}}