A novel filtering algorithm is presented to restore images corrupted b
y impulsive noise. As a preprocessing procedure of the noise cancellat
ion filter, an improved impulse detector is used to generate a binary
flag image, which gives each pixel a flag indicating whether it is an
impulse. This flag image has two uses: (1) a pixel is modified only wh
en it is considered as an impulse; otherwise, it is left unchanged, an
d (2) only the values of the good pixels are employed as useful inform
ation by the noise cancellation filter. To remove noises from the corr
upted image, we propose a new filter called a polynomial approximation
(PA) filter, which is developed by modeling a local region with a pol
ynomial that can best approximate the region under the condition of le
ast squared error. Furthermore, an adaptive approach is introduced to
automatically determine the orders of the polynomials. The proposed tw
o kinds of PA filters, fixed-order and adaptive-order PA filters, are
tested on images corrupted by both fixed-valued and random-valued impu
lsive noise. Major improvements are obtained in comparison with other
state-of-the-art algorithms. (C) 1998 Society of Photo-Optical Instrum
entation Engineers.