Characterizing the composite noise of a camera used as a sensor for position estimation

Abstract

This article presents the characterization of inherent composite noise of a camera based position measurement system. Such an examination is needed to use a camera system as the position measurement sensor for direct implementation in Kalman filter based positional trajectory estimation of moving objects. Time and frequency domain analysis are carried out to characterize the noise in units of measured position values. The qualification of the noise to be White and Gaussian (WGN) makes it possible to use the position measurements for optimal estimation in a Kalman filter like estimation procedure. Further the observation is justified with an experiment of position tracking and subsequent trajectory estimation of a ball thrown under gravity with the help of linear kinematic model.