The background modeling for the video sequences using Radial Basis Function neural network

摘要

This paper an adapted background subtraction system based on Radial Basis Function neural network is given. In the suggested system for each background pixel a two-state machine is considered in which the RBF¹ neural network is used for crossing between different states of the machine. In suggested system is flexible to light changes and very few movements of background object and the presence of new object in the background and other challenges mentioned in the field of estimate the background. The test are down on image group of Wallflower data set, the obtained result confirm the efficiency of the suggested solution.