Abstract:
Visual Monitoring System requires efficient estimation of the background,
background modelling and accurate motion detection. Background subtraction is the
simplest approach to classify the background and foreground. Now a day’s most of the
modern smart video surveillance system requires robust detection of the foregrounds
which effectively handle various environment challenges. Literature shows not a single
algorithm ably handle all constraints, on the basis of the applications researchers
generally, focuses on the classification or segmentation techniques. The objective of this
paper is to provide the rigorous and comprehensive study of the various background
modeling approaches and foreground detection techniques. This survey discussed some of
the popular statistical pixel based parametric and non parametric approaches, region
based approached and some of the popular hybrid approaches for the background
handles. The aim of the study is to handle the various datasets and real time video
sequences challenges. in spite of of all the literature survey still some of the challenges
like object abrupt motion, object complex silhouette, object appearance and camera
motion are challenging issues for the researchers.
