
Correction method for stripe nonuniformity
Weixian Qian,* Qian Chen, Guohua Gu, and Zhiqiang Guan
440 Laboratory, Electronic Engineering and Optoelectronic Techniques,
Nanjing University of Science and Technology, Nanjing, China
*Corresponding author: developer_plus@163.com
Received 17 April 2009; revised 25 January 2010; accepted 23 February 2010;
posted 25 February 2010 (Doc. ID 110022); published 24 March 2010
Stripe nonuniformity is very typical in line infrared focal plane arrays (IR-FPA) and uncooled staring
IR-FPA. In this paper, the mechanism of the stripe nonuniformity is analyzed, and the gray-scale co-
occurrence matrix theory and optimization theory are studied. Through these efforts, the stripe non-
uniformity correction problem is translated into the optimization problem. The goal of the optimization
is to find the minimal energy of the image’s line gradient. After solving the constrained nonlinear opti-
mization equation, the parameters of the stripe nonuniformity correction are obtained and the stripe
nonuniformity correction is achieved. The experiments indicate that this algorithm is effective and
efficient. © 2010 Optical Society of America
OCIS codes: 100.2000, 110.3080.
1. Introduction
Current infrared focal plane arrays (IR-FPA) are fun-
damentally limited by their inability to calibrate out
component variations [1,2]. Fixed pattern noise that
is caused by the nonuniform response of the sen-
sors gives uncorrected images with a white-noise-
degraded appearance. Stripe nonuniformity is a
special kind of nonuniformity, and very typical in
the line IR-FPA and the uncooled staring IR-FPA.
Figure 1 shows an image of an uncooled staring
IR-FPA, and it has the column stripe nonu niformity.
Nonuniformity correction (NUC) techniq ues have
been developed and implemented to perform the
necessary calibration for most IR sensing applica-
tions. These correction techniques can be divided into
two primary categories: 1) reference-based correction
using calibrated images on startup and 2) scene-
based techniques continually recalibrating the sen-
sor for parameter drifts.
The most typical reference-based correction meth-
ods are the so-called “one-point” correction method
and the “two-point” correction method. Problems
with the reference-based methods have been well
documented in the literature [3,4]. A number of re-
searchers have developed scene-based algorithms
for NUC. Ullman and Schechtman studied a simple
gain adjustment algorithm, but their method pro-
vides no mechanism for canceling additive offsets
[5]. Scribner et al. discussed a least-mean-square-
based nonuniformity correction algorithm [6,7], but
it involves complex computations that are not suita-
ble for real-time implementation. Harris and Chiang
introduced the constant-statistics constraint NUC
algorithm [8], which needs many image sequences
for parameter estimation and produces ghosting
artifacts.
All the scene-based NUC algorithms above are de-
signed for the common nonuniformity. To correct the
stripe nonuniformity, special methods should be con-
sidered. The destriping algorithms are often used in
the stripe NUC. The simplest destriping algorithm is
to process the image data with a low-pass filter using
discrete Fourier transform [9]. This method is sim-
ple, but it often does not remove all stripes and leads
to significant blurring within the image. Some re-
searchers remove the stripes using wavelet analysis,
which takes advantage of the scaling and directional
properties to detect and eliminate striping patterns
[10]. Chen et al. [11] proposed a power filtering meth-
od to distinguish the striping-induced frequency
components using the power spectrum, and then
0003-6935/10/101764-10$15.00/0
© 2010 Optical Society of America
1764 APPLIED OPTICS / Vol. 49, No. 10 / 1 April 2010