Touch screen point detection for time-varying inputs

Abstract

Touch screen is widely used as a basic input device for hand-held devices. As it is the main input method, accuracy is an important factor. The accuracy requirement motivates enhanced signal processing for obtaining better decisions on touch points. This thesis proposes a touch screen system model with time-varying inputs. The maximum a posteriori (MAP), maximum likelihood (ML) and Kalman filter methods as well as ML estimation of position-increments are used to estimate touch points. MAP and ML methods are based on estimating the probability of touch positions. The Kalman filter is based on the minimum-mean-squared error criterion. ML estimation of position-increments focuses on tracking the increments of successive position increments. Simulation results show that the proposed methods achieve higher accuracy than the conventional position estimation based on simple interpolation methods.