ARTICLE IN PRESS

Pattern Recognition Letters xxx (2006) xxxxxx

www.elsevier.com/locate/patrec

Image segmentation by histogram thresholding

using hierarchical cluster analysis
Agus Zainal Arin

a,*

, Akira Asano

a
b

Graduate School of Engineering, Hiroshima University, 1-4-1 Kagamiyama, Higashi Hiroshima 739-8527, Japan
Division of Mathematical and Information Sciences, Faculty of Integrated Arts and Sciences, Hiroshima University,
1-7-1 Kagamiyama, Higashi Hiroshima 739-8521, Japan
Received 14 June 2005; received in revised form 22 February 2006

Communicated by G. Borgefors

Abstract
This paper proposes a new method of image thresholding by using cluster organization from the histogram of an image. A new similarity measure proposed is based on inter-class variance of the clusters to be merged and the intra-class variance of the new merged
cluster. Experiments on practical images illustrate the eectiveness of the new method.
 2006 Elsevier B.V. All rights reserved.
Keywords: Image thresholding; Clustering; Inter-class variance; Intra-class variance

1. Introduction
Thresholding is a simple but eective tool for image segmentation. The purpose of this operation is that objects
and background are separated into non-overlapping sets.
In many applications of image processing, the use of binary
images can decrease the computational cost of the succeeding steps compared to using gray-level images. Since image
thresholding is a well-researched eld, there exist many
algorithms for determining an optimal threshold of the
image. A survey of thresholding methods and their applications exists in literature (Chi et al., 1996).
One of the well-known methods is Otsus thresholding
method which utilizes discriminant analysis to nd the
maximum separability of classes (Otsu, 1979). For every
possible threshold value, the method evaluates the goodness of this value if used as the threshold. This evaluation
uses either the heterogeneity of both classes or the homoge*

Corresponding author. Tel./fax: +81 82 424 6476.

E-mail addresses: agusza@hiroshima-u.ac.jp (A.Z. Arin), asano@
mis.hiroshima-u.ac.jp (A. Asano).
0167-8655/$ - see front matter  2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.patrec.2006.02.022

neity of every class. By maximizing the criterion function,

the means of two classes can be separated as far as possible
and the variances in both classes will be as minimal as possible. This method still remains one of the most referenced
thresholding methods.
Minimum error thresholding method nds the optimum
threshold by optimizing the average pixel classication
error rate directly, using either exhaustive search or an
iterative algorithm (Kittler and Illingworth, 1986). This
method assumes that an image is characterized by a
mixture distribution with the population of object and
background classes are normally distributed. The probability density function of the mixture distribution is estimated
based on the histogram of the image.
Since the thresholding problem can be regarded as a
clustering problem, Kwon (2004) proposed a threshold
selection method based on the cluster analysis. He proposed a criterion function that involved not only the histogram of the image but also the information on spatial
distribution of pixels. This criterion function optimizes
intra-class similarity to achieve the most similar class and
inter-class similarity to conrm that every cluster is well

ARTICLE IN PRESS
2

A.Z. Arin, A. Asano / Pattern Recognition Letters xxx (2006) xxxxxx

separated. The similarity function used all pixels in two

clusters as denoted by their coordinates.
Criterion-based thresholding algorithms are simple and
eective for two-level thresholding. However, if a multilevel thresholding is needed, the computational complexity
will exponentially increase and the performance may
become unreliable (Chang and Wang, 1997).
This paper proposes a novel and more eective method
of image thresholding by taking hierarchical cluster organization into account. The proposed method attempts to
develop a dendrogram of gray levels in the histogram of
an image, based on the similarity measure which involves
the inter-class variance of the clusters to be merged and
the intra-class variance of the new merged cluster. The bottomup generation of clusters employing a dendrogram by
the proposed method yields a good separation of the clusters and obtains a robust estimate of the threshold. Such
cluster organization will yield a clear separation between
object and background even for the cases of nearly unimodal or multimodal histogram. Since the proposed method
performs an iterative merging operation, the extension into
multi-level thresholding problem is a straightforward task
by just terminating the grouping when the expected number of clusters of pixel values are obtained.
The concept of developing a dendrogram from a histogram was already employed in our previous work presented in a conference paper (Arin and Asano, 2004).
This paper improves its merging criteria by involving
inter-class variance and intra-class variance in the similarity measurement, so as to maximize the distance of cluster
means, as well as the variance of the new merged cluster.
This paper also provides comparisons of the quality of
binarizations among the proposed method, Otsus method,
and Kwons method, and shows the superiority of our
method to the other methods, based on three kinds of
quantitative comparison. The method has been developed
not only from the theoretical interest, but also a practical
applicability has been already presented in (Arin et al.,
2005).
The rest of this paper is organized as follows: The proposed algorithm is developed in Section 2 its experimental
results and quantitative comparison with other methods

are discussed in Section 3, and nally, conclusions are presented in Section 4.

2. Histogram thresholding
If the histogram of an image includes some peaks, we
can separate it into a number of modes. Each mode is
expected to correspond to a region, and there exists a
threshold at the valley between any two adjacent modes
(Cheng et al., 2002). We propose a novel algorithm for
estimating the optimal threshold using cluster analysis.
Initially, we regard that every non-empty gray level is a
separate mode contained in a cluster. Then the similarities
between adjacent clusters are computed, and the most similar pair is merged. The estimated threshold for the usual
two-level thresholding is obtained by iterating this operation until two groups of gray levels are obtained.
The hierarchical tree of this unication process is best
viewed graphically as a dendrogram. Let us consider an
example gray scale image, which contains 11 11 pixels
and consists of 43 non-empty gray levels ranging from 98
to 198 with the histogram shown in Fig. 1(a). Our method
generates the dendrogram shown in Fig. 1(b). The numbers
along the horizontal axis of dendrogram represent the indices of the gray levels numbered from 1 to 43. The height of
the links between clusters represents the order of merging
operations. The estimated thresholds for the t-level thresholding are obtained by separating the dendrogram into t
groups by cutting the branch. It indicates that the multilevel thresholding is achieved quite straightforwardly by
our method. For the usual two-level thresholding, i.e.,
the usual binarization, the threshold is obtained by cutting
the dendrogram at the highest branch.
2.1. Cluster merging strategy
Let Ck be the kth cluster of gray levels in the ascending
order, and Tk be the highest gray level in the cluster Ck.
Consequently, the cluster Ck contains gray levels in the
range [Tk1 + 1,Tk], if we dene T0 = 1.
The proposed algorithm of cluster merging is summarized as follows:

Fig. 1. (a) Histogram of the sample image and (b) the obtained dendrogram.

ARTICLE IN PRESS
A.Z. Arin, A. Asano / Pattern Recognition Letters xxx (2006) xxxxxx

1. We assume that the target histogram contains K dierent non-empty gray levels. At the beginning of the merging process, each cluster is assigned to each gray level,
i.e. the number of clusters is K and each cluster contains
only one gray level.
2. The following two steps are repeated (K  t) times for
t-level thresholding.
2.1. The distance between every pair of adjacent clusters
is computed. The distance indicates the dissimilarity of the adjacent clusters, and will be dened in
the next subsection.
2.2. The pair of the smallest distance is found, and these
clusters are unied into one cluster. The index of
clusters Ck and Tk are reassigned since the number
of clusters is decreased one by the merging.
3. Finally t clusters, C1, C2, . . ., Ct, are obtained. The gray
levels T1, T2, . . ., Tt1, which are the highest gray levels
of the clusters, are the estimated thresholds. For the
usual two-level thresholding, t = 2 and the estimated
threshold is T1, i.e., the highest gray level of the cluster
of lower brightness.

2.2. Distance measurement

Tk
X
zT k1 1

pz;

K
X

P C k 1.

k1

This function indicates the occurrence probability of pixels

belonging to the cluster Ck.
The distance between the clusters C k1 and C k2 is dened
as
DistC k1 ; C k2 r2I C k1 [ C k2 r2A C k1 [ C k2 .

The two parameters in the denition correspond to the

inter-class variance and the intra-class variance, respectively. The inter-class variance, r2I (C k1 , C k2 ), is the sum
of the square distances between the means of the two clusters and the total mean of both clusters, and dened as
follows:

P C k1
2
mC k1  MC k1 [ C k2 
P C k1 P C k2

P C k2
mC k2  MC k1 [ C k2 2
P C k1 P C k2
P C k1 P C k2

P C k1 P C k2

mC k1  mC k2  ;
3

where m(Ck) denotes the mean of cluster Ck, dened as

follows:
mC k

1
P C k

Tk
X

zpz

zT k1 1

and MC k1 [ C k2 denotes the global mean of the clusters

C k1 and C k2 , dened as follows:
MC k1 [ C k2

P C k1 mC k1 P C k2 mC k2
.
P C k1 P C k2

The intra-class variance, r2A C k1 [ C k2 , is the variance of all

pixel values in the merged cluster, and dened as follows:
r2A C k1 [ C k2

The proposed denition of the distance between two

adjacent clusters in the histogram is based on both the difference between the means of the two clusters and the variance of the resultant cluster by the merging.
To measure the above two characteristics, we regard
the histogram as a probability density function. Let h(z),
z = 0, 1, . . ., L  1, be the histogram of the target image,
where z indicates the gray level and L is the number of
available gray levels including empty ones. The histogram
h(z) indicates the occurrence frequency of the pixel with
gray level z. If we dene p(z) as p(z) = h(z)/N, where N
is the total number of pixels in the image, p(z) is regarded
as the probability of the occurrence of the pixel with gray
level z. We also dene a function P(Ck) of a cluster Ck as
follows:
P C k

r2I C k1 [ C k2

1
P C k1 P C k2


T k2
X

z  MC k1 [ C k2 2 pz.

zT k 1 1 1

3. Experimental results
In order to evaluate the performance of the proposed
method, our algorithm has been tested using images 15
(see Fig. 2). Image 2 and 5 are provided by The MathWorks, Inc. Fig. 3 shows the corresponding histogram of
the original images. It is observed that the shapes of histograms are not only bimodal or nearly bimodal, but also
unimodal or multimodal.
We have compared our method with three others: (i)
Otsus thresholding method (Otsu, 1979), (ii) Minimum
error thresholding method (KIs method) (Kittler and
Illingworth, 1986), and (iii) Kwons threshold selection
method (Kwon, 2004). The thresholded images using the
proposed method are shown in Fig. 4, while Figs. 57 show
thresholded images by using Otsus method, KIs method,
and Kwons method, respectively. The threshold values
determined for each image using the three dierent methods are summarized in Table 1.
The ground truth images shown in Fig. 8 are generated
by manually thresholding the original images shown in
Fig. 2. We can see from Fig. 4 that the proposed method
produces images that are successfully distinguished from
the backgrounds. The results also correspond well to the
images in Fig. 8. Furthermore, whereas some objects or
parts of the objects that are invisible in Figs. 57 become
discernible in Fig. 4. As shown in these gures, more object

ARTICLE IN PRESS
4

A.Z. Arin, A. Asano / Pattern Recognition Letters xxx (2006) xxxxxx

Fig. 2. Original images: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Fig. 3. Histogram of the experimental images: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Fig. 4. Thresholded image obtained by the proposed method: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Fig. 5. Thresholded image obtained by Otsus method: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

ARTICLE IN PRESS
A.Z. Arin, A. Asano / Pattern Recognition Letters xxx (2006) xxxxxx

Fig. 6. Thresholded image obtained by KIs method: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Fig. 7. Thresholded image obtained by Kwons method: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Table 1
Threshold values determined using three threshold selection algorithms
Sample images

Image
Image
Image
Image
Image

1
2
3
4
5

Threshold selection methods

Otsus
method

Kwons
method

KIs
method

The proposed
method

77
125
122
110
96

30
96
174
146
127

44
132
92
56
0

59
105
161
121
102

the method against the two other methods, using the misclassication error (ME), (Sezgin and Sankur, 2004), relative foreground area error (RAE) (Zhang, 1996; Sezgin
and Sankur, 2004), and modied Hausdor distance
(MHD) (Dubuisson and Jain, 1994; Sezgin and Sankur,
2004).
ME is dened in terms of the correlation of the images
with human observation. It corresponds to the ratio of
background pixels wrongly assigned to foreground, and
vice versa. ME can be simply expressed as
ME 1 

pixels are misclassied into background in Figs. 57 than

that in Fig. 4. The border line around the image 5 is classied as the only object in Fig. 6, due to extreme dierence
of frequency from the other peaks in the histogram.
In order to compare the quality of the thresholded
images, we quantitatively evaluated the performance of

jBO \ BT j jF O \ F T j
;
jBO j jF O j

where background and foreground are denoted by BO and

FO for the original image, and by BT and FT for the test
image, respectively. RAE measures the number of discrepancy of thresholded image with respect to the reference image. It is dened as

Fig. 8. The ground-truth of the original images: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

ARTICLE IN PRESS
6

A.Z. Arin, A. Asano / Pattern Recognition Letters xxx (2006) xxxxxx

Table 2
Performance evaluations of the proposed method and comparison with
three other methods
Sample
images

MHDF O ; F T maxd MHD F O ; F T ; d MHD F T ; F O ;

Threshold selection methods

Otsus
method

Kwons
method

KIs
method

The proposed
method

ME
Image
Image
Image
Image
Image

1
2
3
4
5

5.71%
8.51%
6.30%
5.06%
3.60%

26.66%
13.12%
9.72%
26.68%
30.69%

2.72%
10.35%
9.37%
16.64%
15.50%

1.91%
4.68%
1.38%
3.05%
2.89%

RAE
Image
Image
Image
Image
Image

1
2
3
4
5

25.66%
22.05%
18.06%
18.90%
19.47%

54.52%
19.73%
21.78%
50.38%
63.02%

10.68%
27.80%
26.85%
66.70%
87.40%

7.81%
4.65%
3.80%
7.59%
13.99%

MHD
Image
Image
Image
Image
Image

1
2
3
4
5

0.77
0.64
0.54
0.53
0.25

7.93
0.87
0.53
18.30
6.35

1.04
0.84
1.21
42.23
31.50

9
where
d MHD F O ; F T

1 X
min kfO  fT k.
jF O j f 2F fT 2F T
O

0.11
0.18
0.04
0.18
0.18

Note: Least values are bold-faced.

8
A O  AT
>
>
<
AO
RAE
A
 AO
>
T
>
:
AT

sured by the modied Hausdor distance (MHD) method.

MHD is dened as

if AT < AO ;
8
if AT P AO ;

where AO is the area of the reference image, and AT are is

the area of thresholded image. The shape distortion of the
thresholded regions to the ground truth shapes can be mea-

FO and FT denote foreground area pixels in the original

image and the result image, respectively.
Table 2 shows the ME, RAE, and MHD values for the
thresholded image compared to the ground-truth image for
each algorithm. For the ME evaluation, the thresholded
images obtained by the proposed method gave the lowest
ME values indicating the smallest ratio of background
pixels wrongly assigned to the foreground, and vice versa.
Similarly, for the RAE evaluation, the least values were
obtained from the images thresholded by the proposed
method indicating the largest match of the segmented
objects regions. MHD evaluation shows that the smallest
amount of shape distortion is obtained by the proposed
method. Thus, according to three evaluations, the proposed algorithm seems to be the best, having less misclassication error and less relative foreground area error.
To demonstrate the robustness of the proposed method
in the presence of noise, its application to several test
images generated from image 2 by adding increasing quantities of noise with Gaussian distribution, with standard
deviations of 1, 8, 26, 57, and 81, respectively. Each test
image was characterized with SNR obtained by

Fig. 9. Test images obtained by adding noise to image 2: (a) noised image with SNR of 42.7 dB, (b) histogram of (a), (c) thresholded image of (a),
(d) noised image with SNR of 4.5 dB, (e) histogram of (d) and (f) thresholded image of (d).

ARTICLE IN PRESS
A.Z. Arin, A. Asano / Pattern Recognition Letters xxx (2006) xxxxxx
Table 3
Performance evaluations of the proposed method for noise robustness
SNR of test images (in dB)

Performance criteria
ME (%)

RAE (%)

MHD

4.5
6.8
13.3
23.2
42.7

36.58
29.55
19.19
6.91
4.86

30.38
19.92
18.35
15.30
2.05

2.08
1.73
1.19
0.24
0.11

Note: Least values are bold-faced.

XN x XN y

SNR 10 log 4XN x

x1

x1
XN y

y1

I 2 x; y
2

Ix; y  I n x; y
y1

5;

where I(x, y) and In(x, y) denote the gray levels of pixel

(x, y) in the reference and test images, respectively. Nx
and Nx denote the number of columns and rows of the
image, respectively. The corresponding SNR of the test
images are 4.5 dB, 6.8 dB, 13.3 dB, 23.2 dB, and
42.7 dB. Fig. 9(a) and (b) illustrate the test image and
its histogram for SNR of 42.7 dB and Fig. 9(d) and (e)
for SNR of 4.5 dB, respectively. The thresholded images
of the corresponding test images are shown in Fig. 9(c)
and (f), respectively. Both images exhibit a good degree
of similarity with the ground truth of the original image
shown in Fig. 8(b) and illustrate the robustness of the
proposed method in the presence of noise. Table 3 shows
the performance evaluation of the proposed method for
all test images with respect to the ground truth shown
in Fig. 8(b). It is noticed from Table 3 that for the test
image with SNR of 23.2 dB, the ME, RAE, and MHD
evaluations obtained by the proposed method are better
than those of three other methods which are applied to
image 2 without noises.
4. Conclusions
In this paper, we have presented a new gray level thresholding algorithm based on the close relationship between
the image thresholding problem and the cluster analysis.
The key point of this algorithm is that the cluster similarity
measurement is used to control the selection of the threshold value.
The proposed similarity measurement uses two criteria,
the mean cluster distances and variance. These criteria
correspond to the intra-class variance and the inter-class
variance, respectively. The iterative merging ultimately

produces two clusters, from which the optimal threshold

can be estimated.
From the evaluation of the resulting images, we conclude that the proposed thresholding method yields better
images, than those obtained by the widely used Otsus
method, KIs method, and Kwons method. The images
thresholded by the proposed method correspond well to
ground truth images. Evaluation of the application to several test images with dierent level of noises illustrates the
robustness of the proposed method in the presence of
noise. It is obviously straightforward to extend the method
to multi-level thresholding problem by terminating the
grouping process as the expected segment number is
achieved.
Acknowledgement
This research is partially supported by the 2004 research-promoting budget of Hiroshima University.
References
Arin, A.Z., Asano A., 2004. Image Thresholding by Histogram
Segmentation Using Discriminant Analysis. In: Proceedings of IndonesiaJapan Joint Scientic Symposium 2004: IJJSS04, pp. 169174.
Arin, A.Z., Asano, A., Taguchi, A., Nakamoto, T., Ohtsuka, M.,
Tanimoto, K., 2005. Computer-aided system for measuring the
mandibular cortical width on panoramic radiographs in osteoporosis
diagnosis. In: Proceedings of SPIE Medical Imaging 2005Image
Processing Conference, pp. 813821.
Chang, C.-C., Wang, L.-L., 1997. A fast multilevel thresholding method
based on lowpass and highpass lter. Pattern Recognition Letters 18,
14691478.
Cheng, H.D., Jiang, X.H., Wang, J., 2002. Color image segmentation
based on homogram thresholding and region merging. Pattern
Recognition 35, 373393.
Chi, Z., Yan, H., Pham, T., 1996. Fuzzy Algorithms: With applications to
images processing and pattern recognition. World Scientic, Singapore.
Dubuisson, M.P., Jain A.K., 1994. A modied Hausdor distance for
object matching. In: Proceedings of ICPR94, 12th International
Conference on Pattern Recognition, pp. A-566A-569.
Kittler, J., Illingworth, J., 1986. Minimum error thresholding. Pattern
Recognition 19, 4147.
Kwon, S.H., 2004. Threshold selection based on cluster analysis. Pattern
Recognition Letters 25, 10451050.
Otsu, N., 1979. A threshold selection method from gray-level histograms.
IEEE Transactions of Systems, Man, and Cybernetics 9, 6266.
Sezgin, M., Sankur, B., 2004. Survey over image thresholding techniques
and quantitative performance evaluation. Journal of Electronic
Imaging 13 (1), 146165.
Zhang, Y.J., 1996. A survey on evaluation methods for image segmentation. Pattern Recognition 29, 13351346.

Review Paper
Image segmentation by histogram thresholding
using hierarchical cluster analysis
NAMA MAHASISWA

A. PENDAHULUAN
Sebelum membahas lebih jauh mengenai isi paper maka terlebih dahulu akan diterangkan apa
yang dimaksud dengan image segmentation dan thresholding.
Salah satu proses yang penting dalam pengenalan objek yang tersaji secara visual (berbentuk
gambar) adalah segmentasi. Segmentasi objek di dalam citra bertujuan memisahkan wilayah
(region) objek dengan wilayah latar belakang. Selanjutnya, wilayah objek yang telah
tersegmentasi digunakan untuk proses berikutnya (deteksi tepi, pengenalan pola, dan
interpretasi objek).
Metode segmentasi yang umum adalah pengambangan citra (image thresholding). Operasi
pengambangan mensegmentasikan citra menjadi dua wilayah, yaitu wilayah objek dan
wilayah latar belakang. Wilayah objek diset berwarna putih sedangkan sisanya diset berwarna
hitam (atau sebaliknya). Hasil dari operasi pengambangan adalah citra biner yang hanya
mempunyai dua derajat keabuan: hitam dan putih.
Sebelum proses segmentasi, citra mengalami beberapa pemrosesan awal (preprocessing)
untuk memperoleh hasil segmentasi objek yang baik. Pemrosesan awal adalah operasi
pengolahan citra untuk meningkatkan kualitas citra (image enhancement).
B. HISTOGRAM CITRA
Informasi penting mengenai isi citra digital dapat diketahui dengan membuat histogram citra.
Histogram citra adalah grafik yang menggambarkan penyebaran kuantitatif nilai derajat
keabuan (grey level) pixel di dalam (atau bagian tertentu) citra. Misalkan citra digital
memiliki L derajat keabuan, yaitu dari nilai 0 sampai L 1 (misalnya pada citra dengan
kuantisasi derajat keabuan 8-bit, nilai derajat keabuan dari 0 sampai 255). Gambar 1
memperlihatkan contoh sebuah histogram citra, yang dalam hal ini k menyatakan derajat
keabuan dan nk menyatakan jumlah pixel yang memiliki nilai keabuan k.

Gambar 1. Histogram citra

Histogram citra menunjukkan banyak hal tentang kecerahan (brightness) dan kontas
(contrast) dari sebuah gambar. Puncak histogram menunjukkan intensitas pixel yang

menonjol. Lebar dari puncak menunjukkan rentang kontras dari gambar. Citra yang
mempunyai kontras terlalu terang (overexposed) atau terlalu gelap (underexposed) memiliki
histogram yang sempit. Histogramnya terlihat hanya menggunakan setengah dari daerah
derajat keabuan. Citra yang baik memiliki histogram yang mengisi daerah derajat keabuan
secara penuh dengan distribusi yang merata pada setiap derajat keabuan pixel.
Histogram adalah alat bantu yang berharga dalam pekerjaan pengolahan citra baik secara
kualitatif maupun kuantitatif. Histogram berguna antara lain untuk perbaikan kontras dengan
teknik histogram equalization dan memilih nilai ambang untuk melakukan segmentasi objek.
PERBAIKAN KUALITAS CITRA
Perbaikan kualitas citra (image enhancement) merupakan salah satu proses awal dalam
pengolahan citra (image preprocessing). Perbaikan kualitas diperlukan karena seringkali citra
yang diuji mempunyai kualitas yang buruk, misalnya citra mengalami derau (noise) pada saat
pengiriman melalui saluran transmisi, citra terlalu terang/gelap, citra kurang tajam, kabur,
dan sebagainya. Melalui operasi pemrosesan awal inilah kualitas citra diperbaiki sehingga
citra dapat digunakan untuk aplikasi lebih lanjut, misalnya untuk aplikasi pengenalan
(recognition) objek di dalam citra.
Pada paper ini perbaikan kulaitas citra menggunakan operasi pengambangan (thresholding).
SEGMENTASI CITRA
Segmentasi citra bertujuan memisahkan wilayah (region) objek dengan wilayah latar
belakang agar objek di dalam citra mudah dianalisis dalam rangka mengenali objek.
Pengambangan citra (image thresholding) merupakan metode yang paling sederhana untuk
melakukan segementasi. Operasi pengambangan membagi citra menjadi dua wilayah, yaitu
wilayah objek dan wilayah latar belakang. Wilayah objek diset berwarna putih sedangkan
sisanya diset berwarna hitam (atau sebaliknya). Hasil dari operasi pengambangan adalah citra
biner yang hanya mempunyai dua derajat keabuan: hitam dan putih.
Dalam paper ini juga dijelaskan algoritma-algoritma apa saja yang dipakai untuk
mendapatkan image yang optimal yang salah satunya menggunakan model algoritma Otsu.
Model Otsu bekerja dengan menggunakan analisis diskriminan untuk menemukan hal dapat
dipisahkan maksimum kelas-kelas. Untuk setiap thresholding yang mungkin metode ini
mengevaluasi kelebihan dari nilai jika ini digunakan sebagai thresholding. Evaluasi metode
ini menggunakan heterogenitas kelas-kelas dan homogenitas tiap kelas. Untuk
memaksimalkan ukuran fungsi rata-rata dari dua kelas dapat dipisah sejauh mungkin dan
perbedaan-perbedaan di dalam kedua kelas menjadi minimal.
Metode diatas hanya salah satu dari thresholding. Metode lain dengan menemukan kesalahan
minimum dengan menggunakan algoritma iterasi (Kittler dan Illingworth). Metode ini
mengasumsikan bahwa suatu image ditandai oleh distribusi campuran bukan dari
histogramnya. Pengukuran image dapat dilakukan dengan 2 cara yaitu persamaan intra class
dan inter class.
Paper ini mengusulkan suatu novel dan metode lebih efektif dari thresholding dengan
organisasi cluster hirarkis kepada pemakai. Metode yang diusulkan mencoba
mengembangkan suatu dendogram dari histogram tingkat keabu-abuan yang didasarkan pada
ukuran persamaan yang melibatkan perbedaan inter class dan intra class (dapat dilihat pada
figure 1). Paper ini juga menyediakan perbandingan-perbandingan dari mutu binarization
diantara metode yang diusulkan (metode Otsu dan metode Kwon).

Fig. 1. (a) Histogram of the sample image and (b) the obtained dendrogram.

Pada bagian 2 diceritakan konsep histogram thresholding dimana secara garis besar konsep
ini menggambarkan suatu histogram gray scale dari suatu objek dan diambil titik tengahnya
di 149. Dari grafik histogram tersebut dibuatkan dendrogram nya sehingga dihasilkan 2 buah
data yang saling berdekatan dari titik tengah tresholding yang sudah ditentukan sebesar 149.
Pada dasarnya strategi yang dilakukan adalah dengan mengumpulkan cluster pada area yang
memiliki similiarity sama dan distance yang saling berdekatan. Similiarity adalah kemiripan
dari suatu image dan distance adalah jarak dari suatu image. Apabila similiarity nya jauh
maka distance pasti jauh dan begitu sebaliknya jika similiarity dekat maka distance akan
dekat. Sebagai contoh orang Indonesia dengan Malaysia hampir tidak bisa dibedakan karena
jarak negara keduanya berdekatan. Tapi untuk orang Amerika dan Afrika akan berbeda
karena jarak negara keduanya berjauhan.
C. HASIL EKSPERIMEN
Pada bagian 3 dilakukan eksperimen terhadap 5 image (a,b,c,d,e) untuk membuktikan
algoritma tresholding dengan image segmentation. Figure 2 (a,b,c,d,e) adalah image asli yang
diuji. Tampak pada figure 3 menampilkan gambaran histogram dari tiap image yang diuji.
Tidak hanya menunjukkan gambaran histogram yang bimodal tapi juga yang unimodal atau
multimodal. Setiap image mempunyai varian frekuensi yang berbeda-beda dan gambaran itu
mewakili dari berbagai image yang diuji coba. Figure 3 (d dan e) menunjukkan gambaran
frekuensi yang tampak ekstrim dibandingkan dengan figure 3 (a,b,c).

Fig. 2. Original images: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Fig. 3. Histogram of the experimental images: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Hasil uji coba dengan metode thresholding image dapat dilihat pada gambar 4 (a,b,c,d,e).
Dari hasil eksperimen dapat dilihat secara visual suatu gambaran image yang jelas dan detil
dari kelima image yang diuji. Pemisahan antara background dan object terlihat lebih jelas
khususnya untuk figure 4b dan 4c.

Fig. 4. Thresholded image obtained by the proposed method: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Dibandingkan dengan thresholding dengan metode otsu (figure 5) pada image yang sama
didapatkan hasil yang kurang detil untuk figure 4b dan 4c.

Fig. 5. Thresholded image obtained by Otsus method: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Bahkan untuk thresholding image yang dilakukan dengan metode KI (figure 6) didapatkan
hasil yang lebih jelek daripada 2 metode sebelumnya. Khususnya untuk figure 6c, 6d dan 6e
image terlihat kabur dan kurang jelas detilnya.

Fig. 6. Thresholded image obtained by KIs method: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Pada figure 7 thresholding image dilakukan dengan metode Kwon. Hasilnya dapat dilihat
terjadi percampuran antara objek dan background sehingga menyebabkan ada detil yang
hilang seperti pada figure 7a,7b, 7d dan 7e.

Fig. 7. Thresholded image obtained by Kwons method: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Hasil akhir dari keempat metode dapat dilihat pada table 1.

Dari table kita dapat melihat perbandingan nilai untuk thresholding dari 4 metode yang
dipakai di dalam pengujian image. Untuk image 1 hasil terbaik dengan nilai 59, image 2
dengan nilai 105, image 3 dengan nilai 161, image 4 dengan nilai 121 dan image 5 dengan
nilai 102. Dari nilai table terlihat untuk metode yang diusulkan memberikan nilai ME yang
lebih baik dari 3 metode yang lain (otsu, KI, Kwon).
Hasil segmentasi image yang sebenarnya dapat dilihat pada figure 8. Pada gambar terlihat
jelas pemisahan antara background dan object secara jelas.

Fig. 8. The ground-truth of the original images: (a) image 1, (b) image 2, (c) image 3, (d) image 4 and (e) image 5.

Untuk menguji hasil kualitas image diantara 2 metode digunakan suatu teknik pengujian yang
dinamakan misclassification error (ME), relative foreground area error (RAE) dan modified
hausdorff distance (MHD) yang hasilnya dapat dilihat di Tabel 2. Dengan demikian, menurut
tiga evaluasi, yang diusulkan algoritma hasil terbaik akan didapat jika mempunyai lebih
sedikit misclassification error, lebih sedikit relative foreground area error dan lebih sedikit
modified hausdorff distance.

Pada figure 9 diperlihatkan suatu tes yang dilakukan untuk menguji noise dari image 2
(butiran beras) untuk setiap metode yang ada. Hasilnya terlihat dengan memberikan noise
yang berbeda pada image yaitu 42,7 dB dan 4,5 dB memperlihatkan hasil yang lebih baik
untuk noise yang lebih kecil. Hasilnya dapat dilihat pada histogram figure 9b dan 9e.

Fig. 9. Test images obtained by adding noise to image 2: (a) noised image with SNR of 42.7 dB, (b) histogram of (a), (c) thresholded image
of (a),
(d) noised image with SNR of 4.5 dB, (e) histogram of (d) and (f) thresholded image of (d).

Hasil lengkap pengujian dengan noise dapat dilihat pada Tabel 3.

D. KESIMPULAN
1. Pada bagian kesimpulan dapat dikatakan penggunaan metode tresholding dengan
menggunakan hierarchical cluster analysis lebih baik hasilnya untuk setiap image
dibandingkan dengan metode Otsu, Kwan dan KI. Dan ini juga sudah dibuktikan
dengan pengujian noise dan robustness (kekuatan) dari algoritma tersebut.
2. Suatu metode pendekatan peningkatan mutu citra yang terbaik untuk satu
implementasi belum tentu baik untuk implementasi lainnya, sebab karakteristik citra
dapat saling berbeda.
3. Secara subyektif citra yang dihasilkan lebih bagus.
4. Konversi dari citra hitam-putih ke citra biner dilakukan dengan operasi
pengambangan (thresholding). Operasi pengambangan mengelompokkan nilai derajat

keabuan setiap pixel ke dalam 2 kelas, hitam dan putih. Dua pendekatan yang digunakan
dalam operasi pengambangan adalah pengambangan secara global dan pengambangan secara
lokal.

E. KELEBIHAN METODE IMAGE SEGMENTATION DENGAN

HIERARCHICAL CLUSTER ANALYSIS
Secara visual metode ini memiliki hasil akhir yang baik di dalam meminimal kan
kehilangan piksel sehingga gambar yang dihasilkan dapat terlihat detilnya atau
pemisahan antara objek dan background terlihat dengan jelas.
Pengujian metode ini dengan 3 pengujian yaitu ME, RAE dan MHD menunjukkan
hasil yang lebih baik dibandingkan dengan metode Otsu, KI dan Kwan.
Untuk image yang diberi noise metode ini juga memberikan hasil yang lebih baik
dibandingkan dengan 3 algoritma yang lain.

F. KELEMAHAN METODE IMAGE SEGMENTATION DENGAN

HIERARCHICAL CLUSTER ANALYSIS
Secara visual metode diatas dapat memisahkan antara objek dan background
dengan baik tapi untuk detil image masih kurang dibandingkan dengan metode
Otsu.
Dari segi penyajian paper ini tidak menjelaskan secara lengkap / kurang rinci
dalam menyajikan informasi metode lama yang dijadikan acuan untuk membuat
metode baru yang lebih efektif.
Dalam paper ini juga tidak menampilkan gambar histogram dari hasil akhir
pengujian sehingga kita tidak dapat melihat perbedaan yang significant terhadap
metode baru yang diusulkan tersebut.