Modelling of initial reference frame for background subtraction

摘要

Background subtraction techniques are commonly used for separating the objects of interest from its background. Most of the techniques required an initial reference or background frame prior to updating the next sequence during tracking process. In this paper, development of the initial frame reference model is afforded by analysing all examples in term of accuracy and execution time. Initial finding suggests that the median method gives promising result for further investigation in our action recognition research.