Sie sind auf Seite 1von 11

RESEARCH ARTICLE

Journal of Medical Imaging and


Health Informatics

Copyright 2016 American Scientific Publishers


All rights reserved
Printed in the United States of America

Vol. 6, 111, 2016

Noise Based Computation of Decay


Control Parameter in Nonlocal Means
Filter for MRI Restoration
Justin Joseph1 , J. Sivaraman2 , R. Periyasamy3 , and V. R. Simi4
1

Research Scholar, Department of Biomedical Engineering, National Institute of Technology, Raipur 492010, India
2
Department of Biomedical Engineering, Vel Tech Multi Tech, Chennai 600062, India
3
Department of Biomedical Engineering, National Institute of Technology, Raipur 492010, India
4
Department of Computer Science and Engineering, Arunachala College of Engineering, Tamil Nadu 629203, India
The noise suppression and edge preservation offered by Nonlocal Means (NLM) filter depends on the selection
of decay control parameter and radii of search and similarity windows. If the decay control parameter is too
small then little noise will be removed, while if it is too high, the image becomes blurry. Empirical selection of
the filter parameters are always subjective and may be far from optimal. The present study demonstrates an
automated method to compute decay control parameter from the noise estimate and an objective method to
identify the optimum values of operational parameters using modified Pratts Figure of Merit (PFOM) on Magnetic
Resonance (MR) images. The decay control parameter is computed as the product of standard deviation of the
noise and an arbitrary constant, termed as coefficient of noise variance. The radii of search, similarity windows
and coefficient of noise variance which offer maximum noise suppression with minimum edge degradation were
identified on multiple MR images with different textural and noise content. It has been observed that regardless
of the textural and noise content of the MR image, the optimum range of coefficient of noise variance lies
between four and eight, typically at seven. For MR images, the minimum radii of similarity and search windows
should be greater than three, beyond which the radii has no significant influence on the performance of NLM,
provided the radius of search window should be sufficiently larger than radius of similarity window, typically four
times. Since the radii of search, similarity windows and coefficient of noise variance is robust to textural and
noise content of the MR image, the computed decay control parameter depends only on the standard deviation
of the noise. Hence, the proposed NLM is fully noise adaptive. The experimental analysis was performed in
Matlab .

Keywords: Decay Control Parameter, Edge Preserving Filter, Nonlocal Means Filter, Pratts Figure of Merit.

1. INTRODUCTION
Non Local Means (NLM) filter is an edge preserving filter,
widely accepted in medical image restoration because of its detail
preserving and optimum smoothing characteristics. However, the
performance of NLM is prone to the selection of its operational parameters and in general they are selected empirically
through qualitative inspection of the restored image. The operational parameters of NLM include the radii of similarity, search
windows and the decay control parameter. The decay control
parameter has dominant influence on the performance of the filter
than the radii of similarity and search windows.
In NLM, the restored intensity is the weighted sum of the pixel
intensities in the whole image or in a search window of restricted
dimension. When the decay control parameter is very high, all

Author to whom correspondence should be addressed.

J. Med. Imaging Health Inf. Vol. 6, No. 4, 2016

the pixels in the search window will have the same weight with
respect to the pixel to be restored. The restored pixel intensity
thus turns to be approximately the average of the intensities of
the pixels in the search window, leading to a strong smoothing.
When the decay control parameter is very low, only few pixels in
the search window with neighbours are very similar to the neighbourhood of the contextual pixel will have significant weights.
The restored intensity thus tends to be the weighted average of
only few of the pixels in the search window leading to a weak
smoothing and perhaps, noise may be retained in the restored
image. In short, if the decay control parameter (h) is too small
then little noise will be removed while if h is too high, the image
becomes blurry.
There had been investigations112 to find out the optimum values of operational parameters of NLM and to frame adaptive and
image driven methods for their automatic selection. Baudes et al.1

2156-7018/2016/6/001/011

doi:10.1166/jmihi.2016.1780

RESEARCH ARTICLE

J. Med. Imaging Health Inf. 6, 111, 2016

Table I. Optimal h values for different types of MR images for different


values of Rs and (Rsearch = 5).2
Rs
Optimal h

T1
PD
T2

1 ( )

2 ( )

3 ( )

4 ( )

5 ( )

1.30
1.26
1.25

1.22
1.18
1.18

1.18
1.14
1.14

1.18
1.14
1.13

1.14
1.11
1.10

recommended similarity and search windows of radii 7 7 and


21 21, respectively and suggested that the similarity window
of size 7 7 is large enough to be robust to noise and small
enough to take care of fine details. The optimum decay control
parameter h was recommended as 10 for a standard deviation
of gaussian white noise, , added to the true image.
Manjon2 observed the optimum value of decay control
parameter for different values of radius of similarity window
(1 Rs 5) and for different levels of noise, on simulated MR
images of T1, T2 and proton Density (PD) series. To stimulate
Rician noise, zero mean gaussian noise was added to real and
imaginary parts of the simulated MR data and the magnitude
image was computed. It was concluded that there is a linear relation between noise level and optimum value of h. The above
study suggested to use the value of h around 1.2 with RS = 2
and Rsearch = 5 for MR images. The objective function used to
identify the optimum operational parameters was maximum value
of Root Mean Squared Error (RMSE). The optimum values of
the decay control parameter, observed by Manjon for MR images
from different series, corrupted by different levels of noise are
furnished in Table I.
Zhan et al.3 proposed a method to adaptively tune the decay
control parameter locally for each pixel in the image, based on
the Smallest Univalue Segment Assimilating Nucleus (SUSAN)
edge response at the pixel. Avanaki et al.4 introduced a linear
estimation of decay control parameter using corner information
of the noisy image. Wang et al.6 induced blocking artefacts into
multiple images of different information content. Blocking artefacts were induced by compressing each of the images into JPEG
with Quality Score (QS) equal to 10. The images were filtered
with NLM with different decay control parameter values (40
h 180) and observed that optimum value of decay control
parameter depends equally on the quantisation parameter and the
information content. It was concluded that the optimum value of
the decay control parameter which registers maximum Peak Signal to Noise Ratio (PSNR) is different for different images, but
confined to a small range. A rough value of the decay control
parameter was estimated initially, as a function of QS, such that
hinitial = 1/QS/100.2 In JPEG compression, 100 is the maximum value of the QS. Consequently, for a highly compressed
image with minimum QS, hinitial tends to be a large value. Similarly hinitial is minimum for slightly compressed images with large
QS. From the initial estimate, a range of decay control parameter (05hinitial 15hinitial ) was defined. The optimum value of h
was computed by iteratively searching for h which yields the
maximum value of the PSNR, within the identified range.
Sun et al.7 introduced white Gaussian noise into multiple
images at various levels (10  100) and measured the PSNR
after restoration with the NLM having different parameter settings as h2 = 10, h2 = 100 and h2 = 1000. The kernel with
h2 = 10 was found suitable for low level noise and gradually
failed for the standard deviation to the noise greater than 50. The
2

kernel with h2 = 100 was active only for high level noises and
kernel with h2 = 1000 was not suitable for any level of noise
implying that which is nonsensical.
Li et al.8 decomposed the noisy image into wavelet sub-bands
and the coefficients in each sub-band were modelled by Generalized Gaussian Distribution (GGD), whose parameters were
used to estimate the noise variance. The adequate
decay control

parameter setting was approximated as h = 2 by using maximum PSNR as the objective function. An adaptive selection of
the radius of the similarity window inversely proportional to the
local variance was proposed by Alhosainy and Badran.9 Optimum value of decay control parameter for different values of
radius of similarity window (1 RS 18) was observed on multiple test images, corrupted with additive white Gaussian noise
of zero mean and standard deviation equal to 20. The optimum
value of decay control parameter was exponentially decreasing with the increase in radius of similarity window on all test
images. Similarly on a single test, image corrupted at different
noise level (10  30), the optimum value of decay control
parameter was observed for different values of radius of the similarity window.
In contradictory to the exponential decrease of optimum value
of the decay control parameter with respect to the increase in
radius of the similarity window (previous observation), the optimum value of the decay control parameter remained constant for
low noise levels. But maximum possible value of SNR, used as
the objective function does not ensure that the edges are preserved and the noise is fully eliminated and hence the decay
control parameter is optimum.
Zia et al.10 used different sets of operational parameters for
low and high frequency regions in the MR image. Beuchers
morphological gradient operators were used to separate low and
high frequency regions. The optimum parameter settings for these
separate regions were identified empirically. Sticking on to the
observation in Manjon,2 the optimum decay control parameter
h = 12, by fixing h = 14 and h = 11 for radius of the similarity window equal to 1 and 2, the radius of the search window
was continuously varied (3 Rsearch 7) and average improvement in RMSE was observed for different noise level (2 
20) for low and high frequency region separately. Similarly, by
maintaining Rsearch = 5 (Rsearch at which maximum improvement
in RMSE was observed), the radius of the similarity window
was varied (1 RS 4) for (2  20), at h = 11 and h =
14, for low and high frequency regions in the image. The optimum parameter setting identified for high frequency region was
Rsearch = 5, RS = 1 and h = 11 and for low frequency region
the optimum setting was Rsearch = 5, RS = 1 and h = 14. The
literature used two separate kernels for computing the weights
corresponding to the pixels falling in the low and high frequency
regions. But in NLM, conceptually, the Gaussian weighting kernel is meant to offer spatially decaying weights to the grey level
differences between the neighbourhood or similarity window of
the pixel to be restored and a contextual pixel in the search
window. RMSE was the objective function used to identify the
optimum radii of similarity and search windows.
Farmer et al.11 observed that for added white Gaussian noise
( = 30) at h = 2 and radius of the search window equal to
10, the value of the radius of the similarity window equal to 1,
offers the minimum Mean Squared Error (MSE). Similarly, for
h = 2 and the radius of the similarity window equal to 1, the

RESEARCH ARTICLE

J. Med. Imaging Health Inf. 6, 111, 2016

minimum MSE was observed at the radius of the search window


equal to 4. The minimum value MSE was observed at h/ equal
to 1.8 at the radius of the similarity window equal to one and the
radius of the search window equal to 4.
Coupe et al.12 extended the 2D NLM into 3D for the restoration of 3D MR images. The decay control parameter was approximated as a function of the standard deviation of the noise, the
size of cubic similarity window |Ni | and an arbitrary constant
, such that h2 = 2 2 Ni . Assuming the noise to be addictive
Gaussian, its standard deviation was estimated from the mean
square estimate of the pseudo-residuals. With an objective to
make the noise restoration robust to the variation in the size of
cubic similarity window, the L2 norm between the voxels falling
in cubic similarity window around the voxel to be restored and
the contextual voxel in the cubic search window was normalized
by the size of the cubic search window. The value of the arbitrary constant , which yields maximum PSNR was identified for
1  1 at different noise levels (03  021). The values
of , which offer maximum PSNR for low and high levels of
noise was identified to be 0.5 and 1, respectively and the default
value was fixed as unity. The mean accuracy of the noise estimate computed via pseudo residuals was 14% with a standard
deviation of 2%.
Inspite of the diverse opinions on the selection of decay control parameter and empirical observations, efforts to adaptively
estimate it from the noise estimate are not made, except in
Coupe et al.12 Minimum MSE,11 maximum RMSE2 and maximum PSNR69 12 were used as objective functions to identify the
optimum value of decay control parameter. These Image Quality Analysis (IQA) indices measure the similarity between the
original and processed image and do not specifically reflect the
resemblance of the edge content and do not quantify the degree
of noise suppression. These IQA indices either do not comply
with the subjective quality ratings and they are prone to noise
level. The maximum value of PSNR or minimum values of MSE
do not ensure that the edges are well preserved and noise is fully
eliminated. Optimum values of decay control parameter, radii of
the search and similarity windows were identified on simulated
MR images2 10 12 rather than on real time images.
This work was to resolve the implications of the empirical
selection of the operational parameters of NLM. An adaptive
computation of decay control parameter from the standard deviation of the noise and an objective method for selecting the
parameters were proposed. The forthcoming discussions detail
the mathematical formulation of NLM, adaptive computation of
decay control parameter from standard deviation of noise and
PFOM based objective function. Following this, the objective
method for identifying the optimum values of operational parameters of NLM, suitable for MR images is described. The optimum values of the operational parameters identified through this
objective method are also furnished.

2. METHODOLOGY
Many of the image restoration schemes like bilateral filter13 average the similar pixels in the image to estimate the restored pixel
intensity. The similarity may be in terms of the closeness of grey
levels alone and sometimes the spatial proximity too. In contrast to bilateral filter, in NLM the measure of similarity between
pixels is more robust as the regions surrounding the pixels are
compared rather than the individual pixel comparison. Pixels far

from the pixel being filtered are not penalized due to its distance
to the current pixel, as happens with the bilateral filter. The pattern redundancy is not restricted to be local therefore; the scheme
is referred as non-local.
Given the corrupted image X, the restored intensity at any
pixel location Y i j, is the weighted sum of all the pixels in the
corrupted image X is given as
Y i j =

N
M 


W i j m n Xm n

m=1 n=1

i = 1 2     M

and

j = 1 2     N

(1)

Constrained by,
0 W i j m n 1 and

N
M 


W i j m n = 1

m=1 n=1

Where M and N are number of rows and columns in the image.


The weight W i j m n depends on the similarity between
the pixels Xi j and Xm n. The similarity between the current pixel being filtered Xi j and one of the pixels in the image
Xm n refers to the grey level similarity of the squared regions
of radius, Rs , termed as similarity window, centred at Xi j
and Xm n is given as
W ijmn
Rs
e  k=R
GXi +kj +kXm+kn+k2 /h2
s
= M N
Rs
2
2
i=1
j=1 e  k=Rs GXi +kj +kXm+kn+k /h
(2)
i = 1 2     M
, j = 1 2     N
, m = 1 2     M
, n =
1 2     N
. Rs is the radius of the similarity window such that
the size of the similarity window is 2Rs + 1 2Rs + 1. The
decay control parameter h is considered as a linear function of
the standard deviation of the noise as observed by Manjon,2 such
that,
(3)
h = kn
K is an arbitrary constant termed as coefficient of noise variance. The parameter h controls the decay of the exponential
function in Eq. (2) and therefore the decay of the weights, as a
function of similarity between the two regions around the contextual pixels. G is a normalized weighting kernel, similar to
gaussian, which penalizes the grey level difference at the pixels
far from the centre of the similarity window. The kernel offers
comparatively higher weight to the pixels near the centre of the
similarity window and the weight exponentially decays with spatial distance from the centre. However, all the values in the 8connected neighbourhood of the centre of the kernel are kept
the same, to avoid over-weighting effects. The spatial weighting
kernel is given as
GRs i + 1 Rs j + 1 =
i = 1 2     Rs

j
i

1
1 
Rs p=i q=j 2i + 1 2j + 1

and

j = 1 2     Rs

(4)

A simulated kernel for a radius of the similarity window equal


to 3 is depicted in Figure 1. There is a special case in which i = m
and j = n, in Eq. (1), the Euclidian distance between all the pixels in the similarity window is zero in this context. Consequently,
3

RESEARCH ARTICLE

J. Med. Imaging Health Inf. 6, 111, 2016

0.0068 0.0068 0.0068 0.0068 0.0068 0.0068 0.0068

Assuming the noise to be of zero mean and Gaussian distributed,


its standard deviation is computed as,14


0.0068 0.0201 0.0201 0.0201 0.0201 0.0201 0.0068

n =
0.0068 0.0201 0.0572 0.0572 0.0572 0.0201 0.0068
G = 0.0068 0.0201 0.0572 0.0572 0.0572 0.0201 0.0068
0.0068 0.0201 0.0572 0.0572 0.0572 0.0201 0.0068
0.0068 0.0201 0.0201 0.0201 0.0201 0.0201 0.0068
0.0068 0.0068 0.0068 0.0068 0.0068 0.0068 0.0068
Fig. 1. Weighting kernel.

the product of the weighting kernel and the Euclidian distance


between intensities, falling inside the similarity windows, turns
to be zero. In this context, the weighting kernel has no influence
and fails to penalize the grey level difference at pixels far from
the centre of the contextual region. For every pixel in the image,
when the weight is calculated with respect to the other pixels in
the image, the numerator of Eq. (2) equals unity for self similarity so that the other regions with close similarity to the current
region get penalized. To offer maximum priority to the similar
regions in the spatially distinct locations, W i j m n i = m
and j = n is taken as,
W ijmn = maxW ijmn

i = m & j = n

(5)

Replacing each pixel in the corrupted image with the weighted


sum of all the pixels in the image seems to be computatively
inefficient and the viable method is to consider the pixels in a
search window of restricted dimension, rather than the whole
image, so that the Eq. (1) turns to be,
Y i j =

Rsearch

Rsearch

W i j i m j n

m=Rsearch n=Rsearch

Xi m j n

(6)

Rsearch is the radius of the search window. In this study the decay
control parameter (h) is considered as a linear function of the
standard deviation of the noise as demonstrated in Eq. (3). But
devising a noise estimator for computing the noise variance in
highly textured MR images which contain morphological edges
as well is a difficult task. Both the edge and noise pixels are
equally sensitive to second order derivative and both of them
register zero crossings. Usually, it is advisable to estimate the
noise from homogenous regions devoid of edges. Hence, a noise
estimator kernel is designed to be insensitive at high frequency
regions as the difference between two kernels L1 and L2 , each
approximating the Laplacian such that,14
1
H = 2L2 L1  2
1

where L1 = 1 4
0

and L2 = 1/2 0 4
1
4

1
0
1

0
1
0

N
M 

1

Xi j H
2 6M 2N 2 i=1 j=1

(7)

A modified version of PFOM is used in this study to identify


the optimum values of the operational parameters of NLM filter rather than the PSNR. PFOM15 is an index usually used to
compare the performance of different edge detection methods.
Ranjani and Thiruvengadam16 have employed PFOM to measure the edge preservation in Synthetic Aperture Radar (SAR)
images despeckled with Dual Tree Complex Wavelet Transform
(DTCWT), using inter-scale dependence.
In this study, PFOM is used to measure the degree of edge
preservation as a mean to identify the optimum values of operational parameters of NLM. The PFOM between the binary
edge maps of the original and smoothened image can be computed as,15
PFOM =

Ns

1
1
MaxN0  Ns  k=1 1+ d2 k

(8)

where N0 and Ns are the number of dominating or strong edges


present in the original MR image and smoothened image respectively. is an arbitrary penalty parameter which penalizes the
misplaced edge pixels. d2 k is the distance between the kth
edge pixel in the edge map of restored image and the nearest
edge pixel in the edge map of the original image. The distance
d2 i j m n is the Euclidean derived Pythagorean distance
which is computed as,

(9)
di j m n = 1 m2 + j n2
The PFOM quantitatively represents the number of edge pixels
diminished during the restoration and simultaneously the number of misplaced or dislocated edge pixels. The edge maps from
the original MR image and smoothened image are generated via
gradient based thresholding. The gradients of both noisy and
restored images are computed using Sobel discrete derivative
kernel. To generate the binary maps of dominating edges in the
noisy MR image and smoothened image, corresponding gradient
images are thresholded with respect to equal gradient thresholds. It is assumed that the pixels with local gradients above the
mean of local gradients in the original image are true edge pixels
and not noise contributed ones. Binary map of dominating edges
in the original and smoothened image is given as

1 if G0 x y GT

f0 = x y =
0 else
(10)

1 if Gs x y GT

and fs x y =
0 else
where the threshold GT is the mean of local gradients of the original image. X = 1 2    M
and y = 1 2    N
. For an image
of size M N the equation is
GT =

N
M 
1 
G x y
MN x=1 y=1 0

(11)

RESEARCH ARTICLE

J. Med. Imaging Health Inf. 6, 111, 2016

G0 and Gs are the gradient images generated from the original


and smoothened image, respectively. (8)(11) implies,
N0 =

N
M 


f0 x y

and

x=1 y=1

Ns =

N
M 


fs x y

(a)

(b)

(12)

x=1 y=1

The magnitude gradient is computed from gradient along horizontal and vertical directions17 such that,

Gx y = gx2 + gy2
(13)
where the gradient along x and y directions,
f x y
x

and

gy =

f x y
y

Fig. 2.

(a) MR image 1 (b) MR image 2.

(14)

For a smoothing kernel with ideal edge preservation, the number of dominating edge pixels in the smoothened and original
image would be equal, with their spatial locations overlapping.
For a perfect match of maps of dominating edges in the original and smoothened image, the PFOM is unity and a value of
0 indicates abrupt mismatch. In other words a value of PFOM
equal to unity ensures that none of the dominating edge pixels in
the original MR image is degraded or misplaced during smoothing. The commonly used value of arbitrary scaling constant
is 1/9,1820 , 215 and 0.2.21 There is no hard and last constrain for
setting and appreciably > 0 as recommended by Evans and
Liu.21 The value of is set to unity to make the PFOM completely dependent over the number of degraded edge pixels and
the accuracy of the location of the preserved edge pixels during
restoration.

3. RESULTS
The NLM proposed in this study has three operational parameters, coefficient of noise variance, Rs and Rsearch . The optimum
values of these parameters are identified using modified PFOM
as an objective function. MR images which are different in textural and noise content are used to identify the optimum values of
coefficient of noise variance, equally suitable for MR images of
different series and planes. Two specimen MR images are shown
in Figures 2(a and b). The specimens used for the experimental analysis are MR images of pre-operative and biopsy-proven
GBM-edema complex, collected from Hind Labs, Government
Medical College Kottayam, Kerala. The specimen images are of
study, MRS, under T2 FLAIR, GRE, DWI, 1000b ASSET and
T1 FS-ECE series. The specification of MR equipment is; Manufacturer: GE Medical Systems, Model Name: SignaHDxt, Acquisition Type: 2D and 1.5T field strength. All the experimental
analysis is performed in Matlab .
Before employing modified PFOM as the objective function to
identify the optimum values of operational parameters of NLM,
the feasibility of minimum/maximum PSNR as the objective
function is analysed, because majority of the work follow PSNR
based protocol. The variation of PSNR between the noisy and
restored images with respect to coefficient of noise variance for
different values of radius of similarity window (Rs = 3, 5 and 7)
for Rsearch = 9 on the MR image 1 is depicted in Figure 3. The
restored images, gradient of restored images and corresponding
binary edge maps for coefficient of noise variance between 2 and
20 for Rs = 3 and Rsearch = 9 are furnished in Figures 4 to 13.

For each value of coefficient of noise variance, the PSNR and


PFOM between the original and restored images are computed.
The maximum PSNR (51.77) is observed for K = 2 (Fig. 4) and
minimum value of PSNR (44.8) is obtained for K = 20 (Fig. 13).
It can be noted from the restored image and its binary edge
maps in Figures 4(ac) that for the parameter setting offering
maximum PSNR, noise pixels are not fully suppressed. For the
parameter setting offering minimum PSNR (Figs. 13(ac)), noise
pixels are fully suppressed but true edge pixels are lost. Hence,
considering the minimum value of PSNR2 or maximum value
of PSNR69 11 12 as objective criterion for selecting the optimum
values of the operational parameters is not viable. The value of
optimum value of coefficient of noise variance is at the midway
of maximum and minimum values of PSNR. But as the PSNR
do not have a standard range like 01 as PFOM, finding the mid
value of PSNR is difficult.
The variation of PFOM with respect to the variation in coefficient of noise variance for MR image 1 and MR image 2 at
different values of radius of similarity window (Rs = 3, Rs = 5
and Rs = 7) for Rsearch = 7 are illustrated in Figures 14(a, b).
Even though both the images are completely different in their
textural and noise content, the variation of PFOM with respect
to the variation in coefficient of noise variance follows the same
pattern. The sharp fall in the PFOM versus coefficient of noise
variance curve belongs to the region of fast suppression of noise.
The exponential region represents the noise pixels are almost
PSNR versus Coefficient of noise variance
52
Similarity window=3
Similarity Window=5
Similarity Window=7

51

Peak Signal to Noise Ratio

gx =

50
49
48
47
46
45
44

10

12

14

16

18

20

Coefficient of noise variance (K)


Fig. 3.

The variation of PSNR with respect to coefficient of noise variance.

RESEARCH ARTICLE
(a)

J. Med. Imaging Health Inf. 6, 111, 2016

(b)

(c)

Fig. 4. (a) Restored MR image for K = 2, Rs = 3, Rsearch = 9, PFOM = 0889 and PSNR = 5177. (b) Gradient of restored image. (c) Binary edge map of
restored image.

(a)

(b)

(c)

Fig. 5. (a) Restored MR image for K = 4, Rs = 3, Rsearch = 9, PFOM = 0853 and PSNR = 505. (b) Gradient of restored image. (c) Binary edge map of
restored image.

suppressed and only true edge pixels remain. This region is at the
midway between the regions of optimum noise suppression and
maximum edge preservation. This useful region lies in between
the values of coefficient of noise variance 4 and 8 for both MR
images in Figure 2. The restored images, gradient and binary
edge map of denoized images corresponding to this region, are
furnished in Figures 6(ac) and Figures 8(ac). It is also apparent
from Figures 14(a, b) that the radius of similarity window has

(a)

(b)

no significant influence on the degree of edge preservation and


noise suppression.
Variation of PFOM with respect to the radius of similarity window on MR image 1 and MR image 2 for the optimum range
of coefficient of noise variance (k = 4 to k = 8) and Rsearch =
11 are depicted in Figures 15(a, b). From Figures 15(a, b) it is
perspective that the minimum possible value of radius of similarity window is 3 and below which the noise will not be fully

(c)

Fig. 6. (a) Restored MR image for K = 6, Rs = 3, Rsearch = 9, PFOM = 0726 and PSNR = 4944. (b) Gradient of restored image. (c) Binary edge map of
restored image.

RESEARCH ARTICLE

J. Med. Imaging Health Inf. 6, 111, 2016

(a)

(b)

(c)

Fig. 7. (a) Restored MR image for K = 8, Rs = 3, Rsearch = 9, PFOM = 0675 and PSNR = 4848. (b) Gradient of restored image. (c) Binary edge map of
restored image.

(a)

(b)

(c)

Fig. 8. (a) Restored MR image for K = 10, Rs = 3, Rsearch = 9, PFOM = 0638 and PSNR = 4764. (b) Gradient of restored image. (c) Binary edge map of
restored image.

suppressed. Beyond this optimum value of the radius of the similarity window, Rs has no influence on noise suppression or edge
preservation.
Variation of PFOM with respect to the radius of search window on MR image 1 and MR image 2 for the optimum range
of coefficient of noise variance (k = 4 to k = 8) and Rs = 1 are
depicted in Figures 16(a, b). It has been already made evident
that the radius of similarity window should be greater than 3.
But the radius of search window should be always considerably

(a)

(b)

greater than the radius of the similarity window. To examine the


influence of lower values of the radius of search window on noise
suppression and edge preservation, the radius of similarity window is kept at 1 even though its optimum value is three. Similar
to the radius of the similarity window, the minimum value of the
radius of the search window also should be greater than 3, below
which the restoration will be incomplete and noise may retain.
Above the optimum value of the radius of the search window = 3,
the increase in its value do not have much influence on edge

(c)

Fig. 9. (a) Restored MR image for K = 12, Rs = 3, Rsearch = 9, PFOM = 0609 and PSNR = 4691. (b) Gradient of restored image. (c) Binary edge map of
restored image.

RESEARCH ARTICLE
(a)

J. Med. Imaging Health Inf. 6, 111, 2016

(b)

(c)

Fig. 10. (a) Restored MR image for K = 14, Rs = 3, Rsearch = 9, PFOM = 0584 and PSNR = 4628. (b) Gradient of restored image. (c) Binary edge map of
restored image.

(a)

(b)

(c)

Fig. 11. (a) Restored MR image for K = 16, Rs = 3, Rsearch = 9, PFOM = 0564 and PSNR = 4573. (b) Gradient of restored image. (c) Binary edge map of
restored image.

blurring. Hence, it can be concluded that regardless of the information content, texture and noise content, while denoising MR
images Rs and Rsearch should be greater than 3 and the decay
control parameter should be in the range of 48.
The variation of PFOM with respect to the variation in coefficient of noise variance at the optimum values of radii of similarity and search window Rs = 3 and Rsearch = 9 for multiple
MR images with different textural content and noise levels (1 =
0.736, 2 = 0253 and 3 = 0268) is depicted in Figure 16(c).

(a)

(b)

For all the MR images used for the experiment, the optimum
value of coefficient of noise variance which offers maximum
edge preservation and noise supression is with in the range of
48 typically at 7.

4. DISCUSSIONS
Non local means filter being an edge preserving restoration
scheme, while fixing the value its operational parameters,

(c)

Fig. 12. (a) Restored MR image for K = 18, Rs = 3, Rsearch = 9, PFOM = 0546 and PSNR = 4524. (b) Gradient of restored image. (c) Binary edge map of
restored image.

RESEARCH ARTICLE

J. Med. Imaging Health Inf. 6, 111, 2016

(a)

(b)

(c)

Fig. 13. (a) Restored MR image for K = 20, Rs = 3, Rsearch = 9, PFOM = 0529 and PSNR = 448. (b) Gradient of restored image. (c) Binary edge map of
restored image.

(a)

PFOM versus Coefficient of noise variance

(b)

0.9
Similarity window=3
Similarity Window=5
Similarity Window=7

0.8
0.75
0.7
0.65

0.85
0.8
0.75
0.7

0.6

0.65

0.55

0.6

0.5

10

12

14

16

18

Similarity window=3
Similarity Window=5
Similarity Window=7

0.9

Pratts Figure of Merrit

0.85

Pratts Figure of Merrit

PFOM versus Coefficient of noise variance


0.95

20

0.55

Coefficient of noise variance(K)

10

12

14

16

18

20

Coefficient of noise variance(K)

Fig. 14. (a) Variation of PFOM with respect to the variation in coefficient of noise variance on MR image 1. (b) Variation of PFOM with respect to the variation
in coefficient of noise variance on MR image 2.

PFOM versus Radius of similarity window

(a)

(b)

0.84
K=4
K=6
K=8

0.82
0.8
0.78
0.76
0.74
0.72

0.82
0.8
0.78
0.76
0.74

0.7

0.72

0.68

0.7

0.66

K=4
K=6
K=8

0.84

Pratts Figure of Merrit

Pratts Figure of Merrit

PFOM versus Radius of similarity window


0.86

0.68
1

Radius of similarity window

10

11

10

11

Radius of similarity window

Fig. 15. (a) Variation of PFOM with respect to the radius of similarity window on MR image 1. (b) Variation of PFOM with respect to the radius of similarity
window on MR image 2.

RESEARCH ARTICLE
(a)

J. Med. Imaging Health Inf. 6, 111, 2016

PFOM versus Radius of search window

Pratts Figure of Merrit

0.9
K=4
K=6
K=8

0.85

0.8

0.75

0.7

0.65

10

11

Radius of search window


PFOM versus Radius of search window

(b)
0.9

K=4
K=6
K=8

Pratts Figure of Merrit

0.88
0.86
0.84
0.82

5. CONCLUSION

0.8
0.78
0.76
0.74
0.72

10

11

Radius of search window


(c)

PFOM versus Coefficient of noise variance

0.95

Image 1
Image 2
Image 3

0.9

Pratts Figure of Merrit

ensure the best trade-off between these qualitative aspects. The


PSNR and RMSE used as objective functions to compute the
decay control parameter in most of the existing models are no
way related to these aspects and they are indices which account
for the similarity between two images. Even though, a linear
relationship between the noise statistics and the decay control
parameter is established in few models, the selection of the model
parameters are done through trial and error. The use of separate computational models of decay control parameter for high
and low frequency regions in the image is not comprehensive.
The models which iteratively compute the optimum value of the
decay control parameter from a predetermined range certainly
affects the computational feasibility of the filter. The natural texture of the image is disturbed in methods in which the decay
control parameter is locally adaptive, especially when it is varied
with respect to the local variance.
In the proposed model, the model parameters are estimated
objectively by using the modified Pratts Figure of Merit. The
modified PFOM is a quantitative measure of the degree of noise
elimination and the extent to which edges are preserved. The
proposed model is computationally feasible and well preserves
the information content and texture of the image as the relation
is unique over the entire image regions.

0.85
0.8
0.75
0.7
0.65
0.6
0.55

Noise adaptive computation of decay control parameter of NLM


and a method to identify the optimum values of operational
parameters, based on modified PFOM as the objective function
were experimentally demonstrated on MR images. It has been
proved that for MR images, to ensure maximum noise suppression with minimum edge degradation, the minimum radii of similarity and search windows should be greater than three, beyond
which the radii has no significant influence on performance of
NLM, provided the radius of search window should be sufficiently larger than radius of similarity window, typically four
times. In the proposed method, the decay control parameter is
computed as the product of standard deviation of the noise and
an arbitrarary constant, termed as coefficient of noise variance.
Regardless of the textural and noise content of the MR image,
the optimum value of this arbitrarary constant remains the same,
typically at 7. The decay control parameter is insensitive to variation in radii of search and similarity windows. The framework
is demonstrated on realtime MR images rather than simulated
ones. The proposed framework resolves the issues connected with
emperical and subjective selection of operational parameters of
NLM.

0.5
2

10

12

14

16

18

20

Coefficient of noise variance(K)


Fig. 16. (a) Variation of PFOM with respect to the radius of search window
on MR image 1. (b) Variation of PFOM with respect to the radius of search
window on MR image 2. (c) The variation of PFOM with respect to the coefficient of noise variance at the optimum values of the radii of search and
similarity windows (Rsearch = 9 and Rs = 3).

especially the decay control parameter, two separate qualitative


aspects has to be taken into account, which are the degree of
noise elimination and the extent to which edges are preserved.
The optimum selection of the operational parameters should
10

Conflict of Interest
None of the authors have any conflict of interest to declare.

Acknowledgments: One of the authors would like to


thank the fund from the DST-FIST, Govt of India, vide
Ref.: SR/FST/College-189/2013, Dated: 6th August 2014.

References and Notes


1. B. Buades, B. Coll, and J. M. Morel, A non-local algorithm for image denoising,
Proceedings of the IEEE Computer Society Conference on Computer Vision
and Pattern Recognition San Diego, CA, June (2005).

J. Med. Imaging Health Inf. 6, 111, 2016

2. J. V. Manjon, J. C. Caballero, J. J. Lull, G. G. Marti, L. M. Bonmati, and


M. Robles, MRI denoising using non local means. Med. Image. Anal. 12, 514
(2008).
3. Y. Zhan, X. M. Zhang, and M. Y. Ding, SUSAN controlled decay parameter adaption for non-local means image denoising. Electron. Lett. 49, 807
(2013).
4. A. N. Avanaki, A. Diyanat, and S. Sodagari, Optimum parameter estimation
for non-local means image de-noising using corner information, Proceedings
of 9th International Conference on Signal Processing, Beijing, China, October
(2008).
5. M. Ertas, A. Akan, I. Yildirim, and M. Kamasak, Image denoising by using
non-local means and total variation, Proceedings of Signal Processing and
Communications Applications Conference, Trabzon, Turkey, April (2014).
6. C. Wang, J. Zhou, and S. Liu, Adaptive non-local means filter for image deblocking. Signal ProcessImage 28, 522 (2013).
7. Z. Sun, S. Chen, and L. Qiao, A general non-local denoising model using
multi-kernel-induced measures. Pattern Recogn. 47, 1751 (2014).
8. H. Li, W. Hu, Z. Xie, and Y. Yan, Proceedings of Chinese Intelligent Automation
Conference, Lecture Notes in Electrical Engineering, edited by Z. Sun, and
Z. Deng, Springer, Berlin, Heidelberg, China (2013), Vol. 256, pp. 795802.
9. A. M. Alhosainy and E. F. Badran, Adapted non-local means filter using variable window size, Proceedings of the International Conference on Information
Technology, Bhubaneswar, India, December (2009).
10. S. Zia, M. A. Jaffar, A. M. Mirza, and T. S. Choi, Rician noise removal from
MR images using novel adapted selective non-local means filter. Multimed.
Tools and Appl. 72, 1 (2014).

RESEARCH ARTICLE
11. B. Farmer, A. S. Govindarajan, and R. T. Suryaprakash, Investigation
of
the
non-local
means
algorithm,
Available:
wwwpersonal.umich.edu/arunsg/NLMeans.pdf.
12. P. Coupe, P. Yger, S. Prima, P. Hellier, C. Kervrann, and C. Barillot, An optimized blockwise nonlocal means denoising filter for 3-D magnetic resonance
images. IEEE T. Med. Imaging 27, 425 (2008).
13. C. Tomasi and R. Manduchi, Bilateral filtering for gray and color images,
Proceedings of Sixth International Conference on Computer Vision, Bombay,
India, January (1998).
14. J. Immerkear, Fast noise variance estimation. Comput. Vis. Image Und.
64, 300 (1996).
15. W. K. Pratt, Digital Image Processing, Wiley, New York (1978).
16. J. J. Ranjani and S. J. Thiruvengadam, Dual-tree complex wavelet transform
based SAR despeckling using inter-scale dependence. IEEE T. on Geosci.
Remote. 48, 2723 (2010).
17. R. C. Gonzalez and R. E. Woods, Digital Image Processing, Addison-Wesley,
Reading, Mass (1992).
18. S. S. Agaian, K. A. Panetta, S. C. Nercessian, and E. E. Danahy, Boolean
derivatives with application to edge detection for imaging systems. IEEE T.
Syst. Man. Cyb. B 40, 371 (2010).
19. P. Bao, L. Zhang, and X. Wu, Canny edge detection enhancement by scale
multiplication. IEEE T. Pattern Anal. 27, 1485 (2005).
20. P. Trahanias and A. Venetsanopoulos, Vector order statistics operators as
color edge detectors. IEEE T. Syst. Man. Cyb. B 26, 135 (1996).
21. A. N. Evans and X. U. Liu, A morphological gradient approach to color edge
detection. IEEE Trans. on Image Process 15, 1454 (2006).

Received: xx Xxxx xxxx. Accepted: xx Xxxx xxxx.

11

Das könnte Ihnen auch gefallen