Extended Recursive Filtering Estimation of Detector Offset Nonuniformity in Infrared Imaging Systems

In [1] we presented a recursive filter to estimate the detector offset nonuniformity (NU) noise present in infrared (IR) imaging systems. Such a filter was derived considering an estimation time-window short enough so that the offset NU can be regarded as a constant in noise. Since the offset NU is non-stationary, upon the arrival of new blocks of IR data new estimates of the offset NU have to be computed. In this paper, this recursive filter is extended by adding to the so-called intra-block processing (where offset NU is a constant) an inter-block processing algorithm. The inter-block processing algorithm considers the time varying effect in the offset NU and models it as a discrete-time Gauss-Markov random process. So, the extended filter is designed to compensate continuously for the NU when new blocks of frames arrive. In addition, the theoretical analysis of the estimator provides us expressions for selecting the appropriate parameters required by the algorithm. The ability of the method to compensate for the offset NU is demonstrated by employing several mid-wavelength IR video sequences.