Hindawi Publishing Corporation

Computational Intelligence and Neuroscience

Volume 2016, Article ID 5302957, 11 pages
http://dx.doi.org/10.1155/2016/5302957

Research Article
An Effective Color Quantization Method Using Octree-Based
Self-Organizing Maps

Hyun Jun Park,1 Kwang Baek Kim,2 and Eui-Young Cha1

1
Department of Computer Engineering, Pusan National University, Busan 609-735, Republic of Korea
2
Department of Computer Engineering, Silla University, Busan 617-736, Republic of Korea

Correspondence should be addressed to Eui-Young Cha; [email protected]

Received 6 September 2015; Revised 9 November 2015; Accepted 18 November 2015

Academic Editor: Toshihisa Tanaka

Copyright © 2016 Hyun Jun Park et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Color quantization is an essential technique in color image processing, which has been continuously researched. It is often used,
in particular, as preprocessing for many applications. Self-Organizing Map (SOM) color quantization is one of the most effective
methods. However, it is inefficient for obtaining accurate results when it performs quantization with too few colors. In this paper,
we present a more effective color quantization algorithm that reduces the number of colors to a small number by using octree
quantization. This generates more natural results with less difference from the original image. The proposed method is evaluated
by comparing it with well-known quantization methods. The experimental results show that the proposed method is more effective
than other methods when using a small number of colors to quantize the colors. Also, it takes only 71.73% of the processing time
of the conventional SOM method.

1. Introduction Usually, mean absolute error (MAE), mean square error

(MSE), and processing time are used to quantify the perfor-
24-bit images can represent up to 16,777,216 colors. A variety mance of the color quantization results. MAE and MSE are
of colors in an image provides the advantages of having calculated by the following, respectively:
greater abilities of expression and aesthetics. However, a
variety of colors becomes a serious problem for most color
image processing. Therefore, research to represent the various ̂ = 1 𝐻 𝑊 󵄩󵄩 ̂ (ℎ, 𝑤)󵄩󵄩󵄩󵄩 ,
MAE (𝑋, 𝑋) ∑ ∑ 󵄩󵄩󵄩𝑋 (ℎ, 𝑤) − 𝑋 󵄩1 (1)
colors within a limited number of colors, which is called color 𝐻𝑊 ℎ=1 𝑤=1
quantization, has been conducted.
Self-Organizing Map (SOM), which is one of the most
̂ = 1 𝐻 𝑊 󵄩󵄩 ̂ (ℎ, 𝑤)󵄩󵄩󵄩󵄩2 ,
MSE (𝑋, 𝑋) ∑ ∑ 󵄩󵄩𝑋 (ℎ, 𝑤) − 𝑋 (2)
effective color quantization methods, provides excellent 𝐻𝑊 ℎ=1 𝑤=1 󵄩 󵄩2
results [1–5]. However, low-frequency colors in the original
image tend to be excluded during the learning process
because of the characteristics of the SOM learning algorithm ̂ denotes the quantized
where 𝑋 denotes the original image, 𝑋
[6]. In addition, when SOM updates the winner node, it also image, and 𝐻 and 𝑊 denote image height and width,
updates neighbor nodes, which causes loss of the original respectively.
color of the neighbor node. In particular, the loss of original
color increases when SOM uses a small number of nodes to
2. Octree Color Quantization
quantize the image with a small number of colors. Therefore,
to complement these SOM disadvantages, we propose an An octree is one of the image-dependent methods classified
effective color quantization algorithm that is effective even as a hierarchical clustering method. It is faster than partitional
when it uses only a small number of colors by using an octree clustering methods, but its performance is relatively poor [7,
color quantization method. 8].
2 Computational Intelligence and Neuroscience

Original

SOM CQ Octree CQ
MAE 26.3584, MSE 391.4248 MAE 40.9539, MSE 958.5143

Figure 1: Strengths and weaknesses of octree color quantization (number of colors 𝐾 = 32).

An octree is a tree structure with up to eight nodes as However, octree has various colors, like greenish and
children. The octree can represent all colors in an image pinkish, which SOM does not have. This is a strength for
within an eight-level tree because the colors are represented octree for the same reason.
with eight bits. At first, color distribution in the image is In this paper, we present a more effective color quanti-
represented using an octree, which then prunes the nodes zation algorithm that complements the weakness of SOM by
until 𝐾 nodes remain. The palette colors are chosen from the using the strengths of octree, which are fast and offer various
remaining 𝐾 nodes. This method is fast and gives good results colors.
[6, 9].
As mentioned above, octree color quantization generates 3. SOM Color Quantization
the palette using the distribution of colors in the image, but
it does not consider the frequency of color. This means that if SOM color quantization shows excellent results with most
an image is composed of similar colors overall but has many images. The results are very natural and have a low MAE
different low-frequency colors or noise, octree’s results can be and MSE because the SOM learning algorithm spontaneously
very poor. considers the distribution of colors.
Figure 1 shows the strengths and weaknesses of octree However, when SOM color quantization has only a small
color quantization. The original image has two high- number of colors (i.e., the number of colors, 𝐾, becomes
frequency colors overall and many low-frequency colors lower), its disadvantage is apparent. The SOM learning
around them. Yellowish and bluish colors occupy most of the algorithm finds the winner node and updates its weight. At
image. Because SOM quantization considers the distribution that time, neighbor nodes of the winner are also updated.
of color, the generated palette has 21 yellowish colors and 5 This means SOM generates a more natural palette, but also
bluish colors. By contrast, the generated palette from octree it means SOM loses the original color of the neighbor
quantization has 5 yellowish colors and 9 bluish colors. nodes. High-frequency colors are learned many times, so
Because of this, it is not influenced by the distribution of they update the weights many times. This means that high-
colors, and it creates a much greater difference from the frequency colors influence neighbor nodes much more,
original image. This is obviously octree’s weakness. because when the winner is updated, the neighbor nodes are
Computational Intelligence and Neuroscience 3

Original SOM CQ MAE 28.9310, MSE 501.6189

SOM CQ

Octree CQ MAE 56.3035, MSE 1609.2847 Octree CQ

Figure 2: The disadvantage of SOM color quantization (number of colors 𝐾 = 16).

also updated. It becomes more serious when the number of colors in the image. This can induce the results to be more
colors, 𝐾, becomes smaller. If 𝐾 becomes too small, then similar to the original image with reduced loss of the original
the SOM map also becomes so small that the influences on colors. Thus, because the initial weights are fixed, we can
updating the winner become much greater. Therefore, the always get the same results, unlike traditional SOM.
overall colors can become similar and the low-frequency The weight initialization should satisfy the following
colors disappear through updating the weights. conditions to complement the problem above.
Figure 2 shows quantized results with 16 colors from (1) The time to create the map must be short.
SOM [3, 6] and octree [9]. SOM (MAE 28.93, MSE 501.62)
obviously has fewer differences from the original image than (2) The map should be created using the colors in the
octree (MAE 56.30, MSE 1609.28). However, visually, the image.
result of octree looks better because the results from SOM (3) It should be configured with a variety of colors from
lost the sky-blue color, even if octree has almost double the the image as much as possible.
MAE. Octree color quantization satisfies these conditions. It
SOM color quantization is excellent method, but the constructs the octree using the colors in the image, and it
number of colors, 𝐾, is small; then its performance is prunes them until only 𝐾 nodes remain. It does not need
decreased. Therefore, it can be a problem when it is used to much time to create the palette. In addition, it has various
quantize an image with a small number of colors. colors because it does not consider the distribution of color.
Therefore, the proposed method initializes the weights using
4. Octree-Based SOM Color Quantization the palette generated by octree quantization.
Figure 3 shows the overall process of the proposed
We present a more effective color quantization algorithm by method. At first, it builds an octree on the input image and
using octree-based SOM color. It complements the disadvan- prunes the octree until 𝐾 nodes remain to get the 𝐾 number
tages of SOM, so it not only gives visually natural results but of colors for initializing the weights of SOM [9].
also provides low MAE and MSE. After initializing, SOM learning is performed. The pro-
The beginning of the SOM learning algorithm initializes posed method uses a two-dimensional SOM and two dif-
the weights with random values. It causes similar colors ferent learning rates for winner and neighbor nodes. The
to be located anywhere on the palette. If they are high- learning rate for neighbor node 𝛼neighbor is a lot smaller than
frequency colors, then low-frequency colors around them are the learning rate for winner node 𝛼winner in order to reduce
lost. Therefore, we intentionally initialize the weights with the loss of low-frequency colors.
4 Computational Intelligence and Neuroscience

Require: image x = (𝑥1 , 𝑥2 , . . . , 𝑥𝑁 ) as a vector of 𝑁 pixels, number of neurons 𝐾.

Ensure: weight w = (𝑤1 , 𝑤2 , . . . , 𝑤𝐾 )
Build octree on image 𝑥 and prune the nodes until 𝐾 nodes remain.
w ← value of 𝐾 pruned nodes
loop counter 𝑡 ← 1
repeat
d𝑡 ← (𝑥𝑡 , 𝑥2𝑡 , 𝑥3𝑡 , . . . , 𝑥𝑁󸀠 ) {generate subset of training pixels}
𝑁󸀠 ← ‖d𝑡 ‖ {number of training data}
𝛼winner ← adjusted learning rate for winner node
𝛼neighbor ← 𝛼winner /100
for 𝑗 ← 1 to 𝑁󸀠 do
{Determine the winner node}
winner ← argmin‖𝑑𝑡,𝑗 − 𝑤𝑘 ‖2
𝑘=1,...,𝐾
{Update weight of the winner and its neighbors}
𝑤new ← (1 − 𝛼)𝑤old + 𝛼 ⋅ 𝑑𝑡,𝑗
end for
𝑡←𝑡+1
until weights converged

Algorithm 1: Learning algorithm of octree-based SOM color quantization.

Input image Construct octree Prune the octree until K Get K colors
nodes remain

Learn the SOM with

Initialize SOM weights two different learning Generate a palette Quantize the image
with K node colors rates, 𝛼neighbor
and 𝛼winner

Figure 3: Overall process of octree-based SOM color quantization.

The proposed method uses different training data in each 900

796.38
iteration. If SOM learns sequential pixels in an image, it 800
means SOM learns similar pixels repeatedly because adjacent 671
700 620
pixels in an image have similar colors. The training data for 574 601
𝑡th iteration, 𝑑𝑡 , is defined as follows for data sampling: 600 536
499.74 496
500 435
𝑑𝑡 = (𝑥𝑡 , 𝑥2𝑡 , 𝑥3𝑡 , . . . , 𝑥𝑁󸀠 ) . (3) 400 341 370.83
300.53 276.35
300 245.59 263.95 259.02 259.06 258.26
Determining the winner node and updating the weights
200
are performed by traditional SOM learning method as shown
in the following: 100 28.70 24.29 21.83 20.01 19.37 19.09 18.89 18.91 18.88
0
󵄩 󵄩2 1 0.5 0.25 0.1 0.05 0.025 0.01 0.005 0.001
winner = arg min 󵄩󵄩󵄩󵄩𝑑𝑡,𝑗 − 𝑤𝑘 󵄩󵄩󵄩󵄩 , End condition
𝑘=1,...,𝐾
(4)
Avg. MAE
𝑤new = (1 − 𝛼) 𝑤old + 𝛼 ⋅ 𝑑𝑡,𝑗 . Avg. MSE
Avg. time (ms)
Algorithm 1 shows the learning algorithm of octree-based
SOM color quantization. Figure 4: Average MAE, MSE, and processing time of all test cases.
In this paper, it assumes that the weights converge if the
average variation in the weights is lower than 0.025, and it is
used as an end condition. The end condition is experimental 5. Experimental Results
value. Figure 4 shows average MAE, MSE, and processing
time of all test cases. When the end condition is set between The proposed method was tested on a set of ten true-color
0.01 and 0.025, it gives best performances. It shows the end images commonly used to evaluate performance. The images
condition 0.025 is reasonable. are shown in Figure 5, and information about the test images
Computational Intelligence and Neuroscience 5

(a) Lenna (b) Parrots (c) Mona Lisa (d) House (e) Mandrill

(f) Venus (g) Bedroom (h) Scream (i) Old man (j) Dragonfly

Figure 5: Test images.

Table 1: Information on test images. expected, the experimental results show that the proposed
method gives better results than the conventional SOM
Image Resolution Number of colors Color difference method when 𝐾 is small. It is the consequence obtained by
(a) Lenna 512 × 512 148,279 97.60 reducing the learning rate of the neighbors. It minimizes the
(b) Parrots 768 × 512 72,079 126.36 influence on the neighbor by the winner, so the proposed
(c) Mona Lisa 1280 × 1920 125,240 97.14 method has the original colors of the palette. It means the
(d) House 1280 × 1024 363,724 136.87 proposed method is more effective than other algorithms
(e) Mandrill 512 × 512 230,427 118.48 when 𝐾 is small.
(f) Venus 1848 × 1173 396,799 131.67 On the other hand, the disadvantage of SOM is reduced
(g) Bedroom 1280 × 1024 454,673 136.26 when SOM uses a big enough size of map for learning,
(h) Scream 999 × 1362 189,162 108.71
because the influence of the winner is decreased. Therefore,
the performance gap between conventional SOM and the
(i) Old man 321 × 481 33,411 107.84
proposed method also decreases when 𝐾 is large. In some
(j) Dragonfly 481 × 321 41,117 72.24 cases, the proposed method gives a larger MAE and MSE,
because the low learning rate of the neighbor disturbs similar
color grouping on the map.
is in Table 1. Test images (i) and (j) are collected from Berkeley Table 5 shows the processing time of the color quan-
color image database (http://www.eecs.berkeley.edu/Research/ tization methods. The proposed method is faster than the
Projects/CS/vision/bsds). All of the color quantization meth- conventional SOM method because the weights are initialized
ods were tested on an Intel i5-4460 3.2 GHz, 8.0 GB RAM with the palette generated by octree. On the other hand,
machine and were implemented in C++. the conventional SOM method initializes the weights with
The color differences shown in Table 1 were calculated random values. It makes SOM need less time before the
with (5). As this value increases, the image has various colors, weights converge. Therefore, the proposed method requires
and the differences among the colors are greater: only 71.73% of the processing time of the conventional SOM
method.
𝑁 󵄩 󵄩󵄩𝑥 − 𝑥 󵄩󵄩󵄩2
∑𝑁
𝑖=1 ∑𝑗=1 󵄩󵄩 𝑖 𝑗󵄩 󵄩
In Figure 6, the peak signal-to-noise ratio (PSNR) is
Color difference = , for 𝑖 ≠ 𝑗. (5) additionally measured to evaluate quantization quality. PSNR
(𝑁 − 1)2 is a popular measure to evaluate the reconstruction quality
of image compression codecs and thus is used to evaluate
We evaluated the proposed method by comparing it with
color compression quality. A higher PSNR means the color
well-known color quantization algorithms, such as popularity
quantization method has higher quality [13]. PSNR is defined
(POP) [10], octree (OCT) [9], median-cut (MC) [10], 𝐾-
as shown in the following:
means (KM) [11], Adaptive Resonance Theory 2 (ART2) [12],
and Self-Organizing Maps (SOM) [3]. Table 2 explains these 𝐿
color quantization methods. PSNR = 20 × log10 ( ). (6)
√MSE
Tables 3 and 4 show a comparison of MAE and MSE,
respectively. 𝐾 means the number of colors used for quan- Figure 7 shows that the proposed method gives more
tizing an image. The best method is indicated in bold. As similar results to the original image, visually, even if it has
6 Computational Intelligence and Neuroscience

Table 2: Various color quantization methods [1].

Methods Features of methods
(i) One of the simplest methods.
POP [10] (ii) 16 × 16 × 16 color histogram using 4 bits per channel uniform quantization.
(iii) K most-frequent colors in the color histogram are used for quantization.
(i) 32 × 32 × 32 color histogram using 5 bits per channel uniform quantization.
MC [10] (ii) It makes cubes that include all of the histogram.
(iii) It repeatedly splits the cubes that have the greatest number of colors until K cubes are obtained.
(i) Tree structure with up to 8 nodes as children, which can represent all colors in an image within an 8-level
OCT [9] tree.
(ii) Color distribution is represented using octree, which then prunes the nodes until K nodes remain.
(i) It starts with K random clusters.
KM [11, 14, 15] (ii) All of the input data are assigned to the cluster that has the minimum distance within the data.
(iii) The centroid of the cluster is calculated as the average of the assigned data.
(i) Unsupervised learning model.
ART2 [12] (ii) It creates new clusters depending on a vigilance test.
(iii) The palette color is chosen from the centroids of the 𝐾 most-frequent clusters.
(i) Unsupervised learning model.
(ii) One-dimensional self-organizing map with K neurons.
SOM [1–5] (iii) It designates the minimum distance node as the “winner” node and then updates the weights of the winner
node and neighbor nodes.
(iv) It repeats the process until the sum of the weight change is less than a certain threshold.

Avg. MAE Avg. MSE

80 5,000

4,000
60
3,000
40
2,000
20 1,000

0 0
8 16 32 64 128 256 8 16 32 64 128 256

POP ART2 POP ART2

MC SOM MC SOM
OCT PM OCT PM
KM KM

Avg. PSNR Avg. time (ms)

35 5,000
30
4,000
25
20 3,000
15 2,000
10
1,000
5
0 0
8 16 32 64 128 256 8 16 32 64 128 256

POP ART2 POP ART2

MC SOM MC SOM
OCT PM OCT PM
KM KM

Figure 6: Average MAE, MSE, PSNR, and processing time of each method.
Computational Intelligence and Neuroscience 7

Octree SOM Proposed method

MAE 53.24, MAE 29.26, MAE 23.23,

MSE 1474.45 MSE 491.30 MSE 295.22

MAE 32.80, MAE 19.47, MAE 19.81,

MSE 580.59 MSE 210.56 MSE 212.14

MAE 24.76, MAE 17.55, MAE 16.22,

MSE 340.30 MSE 204.07 MSE 166.52

MAE 24.76, MAE 16.19, MAE 16.93,

MSE 386.94 MSE 161.18 MSE 171.55

Figure 7: Resulting images from the color quantization methods (𝐾 = 16).

similar MAE and MSE. Overall colors in the palette generated 6. Conclusions
by SOM become similar if 𝐾 is small. However, the proposed
method has colors more similar to the original image because In this paper, an effective color quantization method is
of the low learning rate of neighbors and initialization by proposed. Having many colors in the original image is a
octree colors. The proposed method gives more similar sky serious problem for image processing, so the number of
color on scream image and gives more similar background colors must be reduced. Therefore, we propose an octree-
color on Mona Lisa image. based SOM color quantization method. It is particularly more
8 Computational Intelligence and Neuroscience

Table 3: MAE comparison of the color quantization methods.

Lenna Parrots
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 41.5 36.6 22.0 16.1 12.7 11.9 101.6 73.5 58.8 22.5 16.5 13.6
MC 31.0 25.1 19.4 15.1 11.5 8.6 64.8 52.1 36.9 25.1 17.8 12.8
OCT 35.9 32.8 23.3 15.5 12.9 9.6 53.9 38.2 28.6 20.3 16.3 12.6
KM 31.1 30.1 24.8 19.7 13.3 12.0 44.7 31.0 26.6 20.2 16.6 12.9
ART2 31.1 22.1 18.5 12.4 10.3 8.5 43.8 35.3 25.9 18.3 14.2 11.3
SOM 27.4 19.5 14.0 10.9 8.6 6.9 42.4 28.4 20.3 14.9 11.5 8.7
PM 27.1 19.8 14.9 10.9 8.8 7.2 41.0 27.8 20.1 15.2 11.8 9.1
Mona Lisa House
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 40.4 28.1 20.4 14.2 12.9 12.5 62.6 52.2 39.1 29.7 21.5 15.8
MC 25.2 20.3 16.1 12.5 9.4 7.4 63.6 54.9 39.6 27.5 20.2 15.9
OCT 29.7 24.8 16.3 13.4 10.4 8.9 51.4 43.3 41.0 25.9 24.8 17.6
KM 30.7 23.5 18.8 14.2 13.6 11.8 43.1 30.8 25.7 22.1 18.5 16.0
ART2 29.2 21.8 16.3 12.3 9.5 7.6 44.1 33.4 29.3 23.6 18.8 15.3
SOM 25.9 17.6 13.3 9.7 7.3 5.6 47.3 33.2 23.7 18.5 14.8 11.8
PM 24.0 16.2 12.0 9.1 7.2 5.7 42.0 30.6 23.8 18.8 14.9 12.0
Mandrill Venus
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 78.6 56.9 40.9 31.0 21.1 15.7 109.6 46.5 33.0 23.8 17.3 13.8
MC 69.3 64.6 59.0 46.2 33.8 21.6 41.8 35.7 33.3 27.5 20.3 15.1
OCT 55.3 51.3 39.0 27.2 22.0 17.9 68.6 56.3 34.9 32.0 22.9 17.8
KM 46.3 34.9 30.3 22.5 18.7 16.8 43.5 34.3 25.5 20.8 17.6 14.7
ART2 50.7 37.3 29.2 24.8 20.0 16.3 44.6 37.4 31.2 22.3 17.9 13.7
SOM 45.3 34.0 26.1 20.5 16.4 13.1 42.5 30.3 21.9 16.9 13.4 10.7
PM 45.0 34.1 26.1 20.5 16.8 13.4 40.9 28.4 21.2 16.7 13.4 10.7
Bedroom Scream
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 102.1 96.5 56.0 36.1 27.1 19.0 59.3 47.3 29.5 20.9 14.6 12.3
MC 74.0 65.8 47.4 37.8 28.4 19.9 54.5 41.0 28.1 20.5 14.2 10.4
OCT 63.8 57.7 47.9 30.6 26.9 18.1 54.6 53.2 42.1 21.9 17.4 13.0
KM 53.8 40.2 33.0 25.9 21.0 17.3 33.6 27.8 24.2 21.7 15.2 13.0
ART2 52.6 47.5 35.6 27.8 21.7 17.7 33.0 26.3 19.8 15.3 12.5 10.3
SOM 53.8 39.2 28.4 21.8 17.2 13.6 40.5 29.3 18.0 13.6 10.6 8.5
PM 51.7 37.3 28.0 21.7 17.2 13.7 33.1 23.2 17.6 13.7 10.7 8.9
Old man Dragonfly
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 71.3 42.6 21.6 13.5 12.9 12.9 39.0 25.8 17.4 14.8 12.8 12.0
MC 27.9 19.0 15.2 11.1 8.8 6.8 26.9 22.7 18.1 13.8 10.9 8.3
OCT 51.6 40.2 17.7 12.0 9.4 7.0 40.2 26.9 22.4 14.3 13.6 9.0
KM 26.8 21.2 18.6 15.8 14.3 12.2 26.5 20.5 20.9 18.3 14.0 13.5
ART2 38.4 27.9 20.9 14.6 11.4 8.5 26.7 23.2 18.1 14.7 11.6 9.3
SOM 22.5 15.9 11.4 8.7 6.5 5.2 21.4 16.2 12.7 10.0 8.9 6.2
PM 22.3 15.7 11.8 8.7 6.8 5.5 21.1 16.9 13.4 10.6 9.6 6.9

effective than other algorithms when the number of colors, 𝐾, quantization method does not consider the distribution of
is small. colors, so its palette includes various colors. Therefore, we use
It uses an octree color quantization method to comple- the palette generated by octree to initialize the SOM weights.
ment the disadvantages of SOM color quantization, which This causes the palette generated by SOM to have more varied
is one of the more effective methods. The octree color colors than conventional SOM. In addition, the learning rate
Computational Intelligence and Neuroscience 9

Table 4: MSE comparison of the color quantization methods.

Lenna Parrots
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 1296.2 1144.6 336.1 175.1 79.8 64.8 8796.1 5976.1 4102.2 362.6 179.5 104.1
MC 577.8 412.4 264.5 163.6 94.8 47.0 2733.2 1716.3 999.7 508.6 274.4 142.9
OCT 692.6 580.6 273.5 126.1 85.4 47.5 1479.2 915.6 469.9 244.6 144.7 85.4
KM 515.3 476.0 334.2 204.9 96.4 77.5 1083.2 534.2 390.6 225.3 150.0 90.7
ART2 517.0 257.9 183.0 82.9 56.0 37.8 1085.8 657.3 353.4 179.7 105.8 68.0
SOM 400.6 210.6 110.0 67.1 42.5 27.8 1027.2 450.2 239.2 133.0 78.8 46.8
PM 396.1 212.1 121.5 66.5 43.1 28.9 960.4 433.1 239.0 135.6 81.0 47.9
Mona Lisa House
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 1330.1 658.7 320.2 103.4 76.5 69.0 2758.1 2044.4 1305.1 844.5 399.1 175.6
MC 496.5 330.9 211.7 131.2 60.3 33.2 2407.0 2003.1 1307.8 674.6 338.8 211.0
OCT 536.3 340.3 155.5 97.4 61.0 43.3 1451.2 1058.8 958.5 370.7 339.3 151.9
KM 597.4 339.1 203.0 117.8 103.5 76.4 1040.3 555.7 381.1 276.0 190.2 137.3
ART2 523.1 275.7 147.8 84.0 48.9 30.2 1137.8 620.0 481.0 306.3 189.7 121.6
SOM 483.0 204.1 118.2 60.4 33.7 20.5 1162.8 598.1 326.9 204.2 128.9 81.7
PM 373.2 164.2 89.0 52.1 32.3 20.4 965.6 530.4 322.4 200.9 126.7 81.7
Mandrill Venus
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 3878.9 2233.5 1200.6 740.0 305.7 151.7 7759.8 1591.6 899.7 450.8 208.9 110.4
MC 3308.9 3054.1 2614.9 1619.9 801.5 347.2 1107.6 884.4 805.7 577.9 303.0 179.2
OCT 1544.7 1358.6 841.2 379.1 245.6 153.7 2290.2 1609.3 600.3 497.9 286.2 150.6
KM 1137.8 647.2 486.3 266.8 183.4 147.6 1038.6 602.2 342.2 231.4 170.8 120.9
ART2 1339.1 733.4 437.0 319.6 206.3 135.4 987.4 691.0 499.0 261.7 166.0 99.8
SOM 1092.9 627.1 358.3 222.7 144.1 93.1 1067.0 577.3 292.5 173.5 107.3 68.5
PM 1078.2 612.4 354.6 221.8 149.7 94.0 905.9 473.3 263.8 163.6 104.7 67.4
Bedroom Scream
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 8107.1 7826.4 2374.7 1061.7 601.6 268.8 2609.4 1859.9 774.7 375.1 128.4 71.8
MC 3429.1 2859.7 1682.0 1121.3 659.6 314.7 1723.1 1149.2 501.3 277.6 133.1 72.8
OCT 2062.4 1753.1 1198.1 480.0 379.3 159.1 1528.1 1474.5 1054.3 256.5 170.2 85.8
KM 1554.1 868.5 588.0 361.6 232.8 156.9 613.1 404.6 312.0 246.7 124.1 88.7
ART2 1465.0 1152.9 647.0 400.9 245.9 158.8 581.6 356.3 204.9 121.3 80.7 54.2
SOM 1591.2 864.1 437.7 258.5 161.0 102.4 936.6 491.3 177.4 102.5 63.0 40.3
PM 1433.4 750.0 423.3 253.0 160.0 101.4 592.1 295.2 166.6 101.8 62.0 42.7
Old man Dragonfly
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 3643.5 1551.8 342.2 88.3 78.7 77.9 1427.9 512.6 230.4 157.5 97.4 66.4
MC 522.2 216.7 138.5 80.4 50.6 26.1 449.2 330.3 234.7 161.5 104.7 59.2
OCT 1432.0 764.2 178.3 75.5 46.1 25.4 990.4 386.9 264.2 105.8 95.2 40.8
KM 386.5 245.3 194.0 138.6 115.9 79.6 387.1 235.7 239.7 183.8 108.4 99.2
ART2 715.5 388.4 245.1 115.6 69.9 38.0 390.7 287.4 175.1 113.3 69.7 44.5
SOM 270.7 138.5 75.4 41.1 24.9 15.8 271.4 161.2 98.9 60.8 43.8 24.3
PM 266.9 135.4 78.5 41.2 25.1 16.3 264.3 171.6 103.1 63.2 50.1 26.6

of the neighbors is set to low to minimize the loss of color The proposed method was evaluated by comparing it
information from updating the winner, because the influence with six well-known quantization methods. MAE, MSE,
of the winner becomes greater when 𝐾 is small. The low and processing time are measured for comparison. The
learning rate of the neighbor means the proposed method can experimental results show that the proposed method is more
have colors similar to the original image. effective than other methods when it uses a smaller number of
10 Computational Intelligence and Neuroscience

Table 5: Processing time (ms) comparison of the color quantization methods.

Lenna Parrots
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 2 2 2 2 3 3 3 3 3 4 4 6
MC 3 2 2 2 2 3 3 3 4 3 4 4
OCT 6 6 7 7 8 8 9 11 10 12 12 13
KM 46 63 97 87 170 323 65 84 124 207 527 665
ART2 31 46 107 88 300 621 68 102 190 360 520 1001
SOM 93 107 139 213 333 548 148 160 207 311 471 825
PM 73 79 97 147 302 387 116 131 162 214 330 581
Mona Lisa House
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 21 19 28 23 21 23 11 11 11 11 12 16
MC 23 21 20 21 21 23 12 12 12 12 11 12
OCT 56 58 65 66 74 75 30 29 32 33 35 44
KM 321 646 778 1179 2246 2074 268 337 540 753 1150 2192
ART2 300 457 524 765 1462 6121 406 676 896 1930 3833 7722
SOM 863 938 1347 1678 2848 4621 485 681 754 1157 1791 2814
PM 612 708 896 1260 1979 3213 359 514 530 721 1135 1904
Mandrill Venus
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 2 2 3 3 3 5 18 18 20 20 19 22
MC 2 3 3 3 3 3 18 18 20 20 18 20
OCT 5 6 6 7 7 8 50 49 57 58 58 58
KM 46 66 96 134 230 441 455 487 740 1133 2065 3458
ART2 62 91 201 408 696 1428 631 760 1503 3408 6930 15597
SOM 108 130 156 220 467 593 953 985 1286 1839 3239 4898
PM 73 83 104 153 332 395 623 755 904 1246 1920 3325
Bedroom Scream
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 12 12 20 11 11 14 10 11 11 20 14 13
MC 10 13 12 11 13 13 13 18 13 13 12 12
OCT 28 28 30 31 33 66 32 31 34 38 37 42
KM 231 295 431 685 1159 2346 261 341 440 701 1183 1975
ART2 338 653 989 2183 4517 9675 179 210 624 1086 2026 4287
SOM 520 825 979 1103 1779 2944 477 688 740 1169 1700 3024
PM 349 549 515 726 1139 2063 413 633 525 841 1214 2054
Old man Dragonfly
𝐾
8 16 32 64 128 256 8 16 32 64 128 256
POP 1 1 1 1 2 2 3 1 1 2 2 2
MC 1 1 1 1 1 1 1 1 1 1 1 2
OCT 3 3 4 4 5 5 3 4 4 4 4 5
KM 19 26 41 39 56 72 19 27 44 68 119 163
ART2 14 16 19 29 43 76 16 19 27 40 72 259
SOM 37 46 78 100 155 257 37 47 64 100 162 277
PM 33 37 52 73 109 195 30 34 47 69 136 194

colors to quantize the colors in the image (𝐾 = 8–32). It gives similar MAE and MSE with conventional SOM, but it is still
lower MAE and MSE. Thus, the proposed method requires faster.
only 71.73% of the processing time of the conventional SOM It is expected that the proposed method will be available
method. On the other hand, when it uses a large number to many applications that have a problem with varied colors
of colors, the performance of the proposed method gives in images.
Computational Intelligence and Neuroscience 11

Conflict of Interests [15] Y.-L. Huang and R.-F. Chang, “A fast finite-state algorithm for
generating RGB palettes of color quantized images,” Journal of
The authors declare that there is no conflict of interests Information Science and Engineering, vol. 20, no. 4, pp. 771–782,
regarding the publication of this paper. 2004.

Acknowledgment
This work was supported by BK21PLUS, Creative Human
Resource Development Program for IT Convergence.

References
[1] C.-H. Chang, P. Xu, R. Xiao, and T. Srikanthan, “New adaptive
color quantization method based on self-organizing maps,”
IEEE Transactions on Neural Networks, vol. 16, no. 1, pp. 237–
249, 2005.
[2] K.-L. Chung, Y.-H. Huang, J.-P. Wang, and M.-S. Cheng,
“Speedup of color palette indexing in self-organization of
Kohonen feature map,” Expert Systems with Applications, vol. 39,
no. 3, pp. 2427–2432, 2012.
[3] A. H. Dekker, “Kohonen neural networks for optimal colour
quantization,” Network: Computation in Neural Systems, vol. 5,
no. 3, pp. 351–367, 1994.
[4] J. Rasti, A. Monadjemi, and A. Vafaei, “Color reduction using
a multi-stage Kohonen self-organizing map with redundant
features,” Expert Systems with Applications, vol. 38, no. 10, pp.
13188–13197, 2011.
[5] Y. Xiao, C.-S. Leung, P.-M. Lam, and T.-Y. Ho, “Self-organizing
map-based color palette for high-dynamic range texture com-
pression,” Neural Computing and Applications, vol. 21, no. 4, pp.
639–647, 2012.
[6] H. J. Park, K. B. Kim, and E. Y. Cha, “An effective color quan-
tization method using color importance-based self-organizing
maps,” Neural Network World, vol. 25, no. 2, pp. 121–137, 2015.
[7] Y. Deng, B. S. Manjunath, C. Kenney, M. S. Moore, and H. Shin,
“An efficient color representation for image retrieval,” IEEE
Transactions on Image Processing, vol. 10, no. 1, pp. 140–147, 2001.
[8] M. E. Celebi, Q. Wen, and S. Hwang, “An effective real-
time color quantization method based on divisive hierarchical
clustering,” Journal of Real-Time Image Processing, vol. 10, no. 2,
pp. 329–344, 2012.
[9] M. Gervautz and W. Purgathofer, “A simple method for color
quantization: octree quantization,” in New Trends in Computer
Graphics, pp. 219–231, Springer, Berlin, Germany, 1988.
[10] P. Heckbert, “Color image quantization for frame buffer dis-
play,” ACM SIGGRAPH Computer Graphics, vol. 16, no. 3, pp.
297–307, 1982.
[11] Y.-C. Hu and M.-G. Lee, “K-means-based color palette design
scheme with the use of stable flags,” Journal of Electronic
Imaging, vol. 16, no. 3, Article ID 033003, 2007.
[12] K. B. Kim, M. Kim, and Y. W. Woo, “Recognition of shipping
container identifiers using ART2-based quantization and a
refined RBF network,” in Adaptive and Natural Computing
Algorithms, pp. 572–581, Springer, Berlin, Germany, 2007.
[13] X. D. Yue, D. Q. Miao, L. B. Cao, Q. Wu, and Y. F. Chen,
“An efficient color quantization based on generic roughness
measure,” Pattern Recognition, vol. 47, no. 4, pp. 1777–1789, 2014.
[14] Y.-C. Hu and B.-H. Su, “Accelerated k-means clustering algo-
rithm for colour image quantization,” The Imaging Science
Journal, vol. 56, no. 1, pp. 29–40, 2008.