COMPARATIVE
STUDY OF BACKGROUND ESTIMATION ALGORITHMS IN VIDEOS
S. A. Kuzmin
Saint-Petersburg State University of Aerospace Instrumentation,
Faculty of
radioengineering, electronics and communications, Chair of
electronic and television systems
Received
March
23, 2012
Abstarct: In first part of article models of statistical
presentation of several types of images as mixture of distributions are studied.
In the second part of the article the difference between MSE values for
histograms of difference images is interpreted as difference in conditions of
acquisition of etalon difference images that gives an ability to a priori make
a conclusion on characteristics of object detection algorithm. Final part
consists of experimental comparison of several background estimation algorithms.
Keywords: background subtraction, background
estimation, inter-frame difference, mixture of distributions, signal-to-noise
ratio.