Nonlinear Lag Correction Based on the Autoregressive Model for Dynamic Flat-Panel Detectors

Abstract

Lag signal problems occur in acquiring X-ray image sequences from dynamic flat-panel detectors due to amorphous pixel photodiodes and incomplete readouts. Based on a linear, time-invariant system with a multiple exponential moving average model for the lag signal, Hsieh et al. proposed a recursive deconvolution algorithm for lag corrections, and Starman et al. proposed a nonlinear correction algorithm to cope with nonlinear lag properties. In this paper, we consider an autoregressive model of order 1 to describe lag signals and conduct lag corrections through simple linear and nonlinear decorrelation schemes for the exposure-dependent lag signals. In order to correct the current image frame, using only the previous frame is enough for the autoregressive model. We also evaluate the lag correction performance of the proposed lag correction algorithms by measuring the lag correction factor to show the successful removal of the lag signals with low computational complexities.