You are on page 1of 8

International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)

Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com


Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 162


Abstract: This paper presents a new modified image
compression algorithm using the color filter array. In the
proposed algorithm first of all a full color image is converted
into Bayer layer i.e. a 2D image with alternating colors. But
as decompression is done by using interpolation techniques
therefore may result in color artefacts and corrupted pixels.
The survey has shown that the compression algorithm based
on Bayer layer suffers from color artefacts so it is required to
reduce them. So this work has proposed an modified
algorithm with color normalization to enhance the results
further. The proposed algorithm is designed and implemented
in MATLAB using image processing toolbox. The
comparisons has shown that the proposed algorithms provides
provides quite significant results.

Keywords: CHANNEL INTERPOLATION, COMPRESSION,
BAYER LAYER, CHANNEL NORMALIZATION.

1. INTRODUCTION
A Demosaicing [1]- [4] is a digital image process used to
reconstruct full color image from incomplete color
samples output from image sensor overlaid with a color
filter array (CFA). It is also called CFA interpolation.
Usage of digital cameras is spreading widely as they are
easy image input devices. The increasing popularity of
digital cameras has provided motivation to improve all
elements of the digital photography signal chain. To
lower cost, digital color cameras typically use a single
image detector. Color imaging with a single detector
requires the use of a Color Filter Array (CFA) [2] which
covers the detector array. In this arrangement each pixel
in the detector samples the intensity of just one of the
many-color separations. The recovery of full-color images
from a CFA-based detector requires a method of
calculating values of the other color separations at each
pixel. These methods are commonly referred as color
interpolation or color demosaicing algorithms. In a
single-detector camera [3], varying intensities of light are
measured at an rectangular grid of image sensors.
To reconstruct a full color image from the data collected
by the color filtering array, a form of interpolation is
needed to fill in the blanks. The mathematics here is
subject to individual implementation, and is called
demosaicing.In this example, we use Adobe Photoshop's
bicubic interpolation to simulate the circuitry of a Bayer
filter device such as a digital camera. The image below
simulates the output from a Bayer filtered image sensor;
each pixel has only a red, green or blue component. The
corresponding original image is shown alongside the
demosaiced reconstruction.

2. COLOR FILTER ARRAY
A color filter array is a mosaic of color filters in front of
the image sensor. Commercially, the most commonly
used CFA configuration is the Bayer filter illustrated
here. This has alternating red (R) and green (G) filters for
odd rows and alternating green (G) and blue (B) filters
for even rows. There are twice as many green filters as
red or blue ones, catering to the human eye's higher
sensitivity to green light.Since the color subsampling of a
CFA by its nature results in aliasing, an optical anti-
aliasing filter is typically placed in the optical path
between the image sensor and the lens to reduce the false
color artifacts (chromatic aliases) introduced by
interpolation. Since each pixel of the sensor is behind a
color filter, the output is an array of pixel values, each
indicating a raw intensity of one of the three filter colors.
Thus, an algorithm is needed to estimate for each pixel
the color levels for all color components, rather than a
single component.
A Bayer filter mosaic is a color filter array (CFA) for
arranging RGB color filters on a square grid of
photosensors. Its particular arrangement of color filters is
used in most single-chip digital image sensors used in
digital cameras, camcorders, and scanners to create a
color image. The filter pattern is 50% green, 25% red and
25% blue, hence is also called RGBG GRGB, or RGGB.
The raw output of Bayer-filter cameras is referred to as
a Bayer pattern image. Since each pixel is filtered to
record only one of three colors, the data from each pixel
cannot fully determine color on its own. To obtain a full-
color image, various Demosaicing algorithms can be used
to interpolate a set of complete red, green, and blue values
for each point.


Figure 1: Bayer Pattern Array ([11])

3. ALGORITHMS
3.1 Nearest Neighbor Replication
Enhanced Image Compression using
modified Color Filter Array

Sanjeev Singla
1
,
Abhilasha
2

1&2
GZS PTU Campus, CSE Deptt. Bathinda, Punjab
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com
Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 163


In this interpolation method [2] - [5], each interpolated
output pixel is assigned the value of the nearest pixel in
the input image. The nearest neighbor can be any one of
the upper, lower, left and right pixels. The algorithm is
nonadaptive, i.e. operates in the same way for all pixel
of the same components in every images. The value of
each interpolated pixel is obtained by copying the value of
the nearest pixel in the Bayer pattern image. This method
introduces significant color errors, especially along edges.
However, since no calculations are performed, this
method may be beneficial in applications where speed is
critical, such as video imaging systems.
An example is illustrated below in Figure 2 for 3x3
block in green plane. Here we assume


Figure 2: Interpolation Demosaicing ([11])

Interpolation of green pixels: the average of the
upper, lower, left and right pixel values is assigned as
the G value of the interpolated pixel. For example :
G8 =(G3+G7+G9+G13) / 4.
Interpolation of red/blue pixels:
Interpolation of a red/blue pixel at a green position: the
average of two adjacent pixel values in corresponding
color is assigned to the interpolated pixel. For example :
B7 =(B6+B8) / 2 ; R7 =(R2+R12) / 2
Interpolation of a red/blue pixel at a blue/red position: the
average of four adjacent diagonal pixel values is assigned
to the interpolated pixel. For example :
R8 =(R2+R4+R12+R14) / 4 ; B12 =(B6+B8+B16+B18)
/ 4

3.2 Bilinear Interpolation
The red value of a non-red pixel is computed as the
average of the two or four adjacent red pixels, and
similarly for blue and green. bilinear interpolation
generates signcant artifacts, especially across edges and
other high-frequency content, since it doesn`t take into
account the correlation among the RGB values [2].


Figure 3: Nearest neighbor output Image (adaptive from
[11])

Figure 4: Bilinear Interpolation Output Image ([11])

3.2 Edge Strength Filter
Edge strength filter that provides local, orientation free
luminance transition information.It observes constant
color difference assumption tends to fail across edges.
The points at which image brightness change sharply are
typically organized into a set of curved line segments
termed edges. By applying the filter to all available
pixels, we get the edge strength map of the input image.
Note that, although the filter result for a singlepixel does
not provide any edge direction information, the
relationship between neighboring pixel results yields the
edge orientation in that neighborhood. However, a
mosaicked image only has one of the three color channels
available for every pixel location and it certainly does not
have complete luminance information at any pixel. That
is why, the edge strength filter can only be applied to a
mosaicked image by making an approximation. A simple
edge strength filter and applied it to the CFA
interpolation problem. The edge strength filter helped us
identify the regions where constant color difference
assumption is likely to fail which in turn lead to improved
demosaicing performance.


Figure 5: Edge strength filter output image (adaptive
from [11])

4. LITERATURE SURVEY
4.1 Joint Demosaicing and sub-pixel based down-
sampling.
Fang et al. (2012) [1] has proposed a fast frequency-
domain analysis approach for joint demosaicing and
subpixel-based down-sampling of single sensor Bayer
images. In this, we integrate demosaicing into down-
sampling by directly performing sub-pixel based down-
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com
Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 164


sampling in the bayer domain, due to which the
computational complexity is reduced.
4.2 Image Demosaicing by Non-Local Similarity and
Local Corelation.
Gang and Chang (2012) [2] has presented an adaptive
demosaicing algorithm by exploiting both the non-local
similarity and the local correlation in the color filter array
image. First, the most flattest nonlocal image patches are
searched in the searching window centered on the
estimated pixel. Second, the patch, which is the most
similar to the current patch, is selected among the most
smoothest nonlocal patches. Third, according to the
similar degree and the local correlation degree, the
obtained nonlocal image patch and the current patch are
adaptively chosen to estimate the missing color samples.
4.3 An Efficient Demosaicing Technique Using
Geometrical Information.
Maalouf et al. (2012) [3] has discussed a bandelet-based
demosaicing method for color images. We have used a
spatial mul-tiplexing model of color in order to obtain the
luminance and the chrominance components of the
acquired image. Then, a luminance filter is used to
reconstruct the luminance component. Thereafter, based
on the concept of maximal gradient of multivalued
images, an extension of the bandelet representation for
the case of multivalued images has proposed. Finally,
demosaicing is performed by merging the luminance and
each of the chrominance components in the multivalued
bandelet transform domain.
4.4 Enhanced Integrated Gradient and its
Applications
Chung et al. (2012) [4] has presented an effective
decision-based demosaicing algorithm for Bayer images .
An enhanced edge-sensing measure called enhanced
integrated gradient can support more gradient
information from various color intensity and color
difference planes under the directional compatibility
constraint. An adaptive green plane enhancement which
works with the enhanced integrated gradient is also
proposed to further improve the efficiency of the
algorithm.
4.5 A Unified Approach To bad Pixel Removal
Jimmy Li et al. (2012) [5] has proposed an adaptive
method to avoid inclusion of those bad pixels in the
interpolation process has proposed. Only defective pixels
which are considerably different from their surrounding
neighbors will be deemed as bad and will be corrected.
This is achieved by adaptively varying the order of
interpolation so that the length of interpolation is shorter
if a bad pixel is located closer to the pixel being
interpolated. The bad pixels are located by using a
median-based multi-shell filter structure.
4.6 New image Quality Assessment Algorithms for
CFA Demosaicing
Maschal and Susan Young (2012) [6] has proposed two
new no-reference quality assessment algorithms. These
algorithms provide a relative comparison of two
demosaicing algorithms by measuring the presence of two
common artifacts, zippering and false coloring, intheir
output images. The first algorithm, the edge slope
measure,tests the overall sharpness of each of the three
color channels,thus estimating the relative edge
reconstruction accuracy of each demosaicing algorithm.
The second algorithm, the false color measure, estimates
deviations from the established constant color difference
image model and performs on green-red and green-blue
color difference planes, therefore estimating the red and
blue channel reconstruction of each demosaicing
algorithm. We evaluate and rank common demosaicing
algorithms using these new algorithms.
4.7 Two-layer color Filter Array For High Quality
Images
Dongjae Lee and Byung-JoonBaekone (2012) [7] has
proposed two-layer color filter array, which has a full
resolution in G channel and a half resolution in R/B
channel. To evaluate the performance, simple demosaic
algorithms are presented. A two-layer color filter array to
obtain a high quality image and demosaic algorithms to
interpolate the proposed two-layer color filter array.
Although demosaic and color filter array design methods
have been improved together, recent research results have
shown the performance saturation since the sub-sampling
of color channels is inevitable. To overcome this
limitation, multilayer color filter arrays have been
developed to obtain two or three color data at single pixel
position.
4.8 Edge Oriented Directional Color Filter Array
Interpolation
Pekkucuksen et al. (2012) [8] has presented a simple edge
strength filter to interpolate the missing color values
adaptively. While the filter is readily applicable to the
Bayer mosaic pattern, we argue that the same idea could
be extended to other mosaic patterns and describe its
application to the Lukac mosaic pattern. Clearly
outperforms other available solutions in terms of CPSNR.
We believe that this edge oriented, directional approach
could prove to be useful for other CFA patterns as well.
4.9 A method of color filter array interpolation with
alias cancellation properties
Glotzbach et al. [9] said that the green image is used to
add high-frequency information and reduce aliasing in
the red and blue images. First, the red and blue images
are interpolated with a rectangular low pass filter
according to the rectangular sampling grid. This fills in
the missing values in the grid, but allows aliasing
distortions into the red and blue output images. These
output images are also missing the high-frequency
components needed to produce a sharp image. However,
because the green image is sampled at a higher rate, the
high-frequency information can be taken from the green
image to improve an initial interpolation of the red and
blue images. A horizontal high pass filter and a vertical
high pass filter are applied to the green image. This
provides the high-frequency information that the low
sampling rate of the red and blue images cannot preserve.

International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com
Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 165


5. PROPOSED ALGORITHM
Step I: In this interpolation method, each interpolated
output pixel is assigned the value of the nearest pixel in
the input image. The nearest neighbor can be any one of
the upper, lower, left and right pixels.
Step II: Interpolation of green pixels: the average of the
upper, lower, left and right pixel values is assigned as the
G value of the interpolated pixel. For example: G8 =
(G3+G7+G9+G13) / 4
Step III: Interpolation of red/blue pixels:
o Interpolation of a red/blue pixel at a green
position: the average of two adjacent pixel values
in corresponding color is assigned to the
interpolated pixel. For example : B7 =(B6+B8) / 2
; R7 =(R2+R12) / 2
o Interpolation of a red/blue pixel at a blue/red
position: the average of four adjacent diagonal
pixel values is assigned to the interpolated pixel.
For example : R8 =(R2+R4+R12+R14) / 4 ; B12
=(B6+B8+B16+B18) / 4
Step IV: The red value of a non-red pixel is computed as
the average of the two or four adjacent red pixels, and
similarly for blue and green.
Step V: Apply color normalization to reduce the color
artefacts from the output image.

6. EXERIMENTAL SET-UP
The proposed algorithm is designed and implemented in
MATLAB using image processing and data analysis
toolbox. For the cross validation we have also
implemented the existing algorithm proposed by Xu et al.
(2012) [13]. 20 different images are also taken as shown
in Table 1 for the experimental purpose.
Table 1 Images taken for experimental analysis.
Table1: Experimental images
Sr. No. Image Name Format Size
1 A .bmp 93
2 B .bmp 148
3 C .bmp 337
4 E .bmp 531
5 F .bmp 721
6 G .bmp 769
7 H .bmp 779
8 I .bmp 1090
9 J .bmp 1153
10 K .bmp 1216
11 l .bmp 2176
12 m .bmp 2204
13 n .bmp 2305
14 o .bmp 2305
15 p .bmp 5626
16 q .bmp 792

6. EXPERIMENTAL RESULTS
This section contains the result for 1
st
.image given Table
1.
Figure 6 has shown the input image for compression.
Figure 6: Input image for compression.

Figure 7 has shown the compressed image as the
compression is done in the same way for both algorithms
so the compressed image is identical in both cases.
Figure 7: The Compressed image
Figure 8 has shown the decompressed image using the
existing algorithm.
Figure 8: The decompressed image using the existing
algorithm.
Figure 9 has shown the decompressed image using the
proposed algorithm.
Figure 9:The decompressed image using the proposed
algorithm.
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com
Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 166


8. PERFORMANCE EVALUATION
Table 2: Original V/s Compression size
Sr.
No.
Imag
e
Name
Original
Image Size
(KB)
Compressed
Image (KB)
1 A 93 32
2 B 148 51
3 C 337 114
4 E 531 178
5 F 721 242
6 G 769 258
7 H 779 261
8 I 1090 365
9 J 1153 386
10 K 1216 407
11 L 2176 727
12 M 2204 737
13 N 2305 770
14 O 2305 770
15 P 5626 1877
16 Q 792 265


Figure 10: Original V/s Compression size

Table 2 and Figure 11 are showing the comparative
analysis of the Peak Signal to Noise Ratio (PSNR). PSNR
contains the power of the input signal over the noisy
signal; i.e. the ration of the noise and the ratio of the
input signal. As it is required by us is to increase the ratio
of the signal over the ratio of the noise. Therefore PSNR
need to be maximized; so our goal is to increase PSNR as
much as possible. Table 2 and Figure 11 have clearly
shown that PSNR is quite more in case of proposed
algorithm than the available algorithm therefore it has
proven that the proposed algorithm is quite effective for
the noisy and corrupted images.

Table 3: PSNR Evaluation
Sr. No. Imag
e
Name
Old Proposed
1 a 58.091 63.7966
9
2 b 45.208
2
53.2158
3 c 60.844
6
62.4577
4 e 59.740
1
65.1356
5 f 75.433
8
78.1988
6 g 69.817
2
76.2616
7 h 44.329
7
53.2772
8 i 86.719
2
96.2612
9 j 66.158
6
72.7398
10 k 53.982
7
63.7966
11 l 96.261
6
99.0000
12 m 52.125
1
62.4577
13 n 60.413
8
69.4132
14 o 59.870
7
68.6574
15 p 71.652
6
82.2822
16 q 79.359
6
86.7192


Figure 11: PSNR evaluations

Table 4 and Figure 12 are showing the comparative
analysis of the Mean absoluter error. As Mean absoluter
error to be minimized; so our goal is to reduce the Mean
absoluter error as much as possible. Table 4 and Figure
12 are clearly shown that Mean absoluter error is less in
the case of the proposed algorithm over the available
technique. Therefore proposed algorithm is providing
better results.
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com
Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 167


Table 4: Mean absoluter error
Sr.
No.
Imag
e
Name
Old Proposed
1 a 0.0453 0.0346
2 b 0.0947 0.0612
3 c 0.0245 0.0239
4 e 0.0537 0.0428
5 f 0.0237 0.0213
6 g 0.0224 0.0170
7 h 0.1480 0.0876
8 i 0.0072 0.0049
9 j 0.0277 0.0221
10 k 0.0563 0.0373
11 l 0.0058 0.0049
12 M 0.0735 0.0411
13 N 0.0271 0.0258
14 O 0.0283 0.0205
15 P 0.0193 0.0122
16 Q 0.0151 0.0103


Figure 12: Mean absoluter error

Table 5 and Figure 13 are showing the comparative
analysis of the Mean Difference. As Mean Difference
needs to be minimized; so our goal is to reduce the Mean
Difference as much as possible. It is clearly shown in
Table 5 and Figure 13 are that the proposed algorithms
has successfully achieve the better results for mean
difference also.
Table 5: Mean difference
Sr.
No.
Imag
e
Name
Old Proposed
1 a 81.5732 42.6580
2 b 357.6827 142.1067
3 c 59.9136 49.3408
4 e 67.7870 26.9418
5 f 11.9999 8.1345
6 g 21.2121 10.1777
7 h 395.8513 141.1839
8 i 3.5184 1.3738
9 j 32.7080 15.5931
10 k 130.4326 42.7576
11 l 1.2541 0.7681
12 m 161.5668 49.1707
13 n 62.1040 22.8695
14 o 66.5978 24.0088
15 p 17.0140 5.9700
16 q 7.9985 3.3435


Figure 13: Mean difference evaluations

Table 6 and Figure 14 are showing the comparative
analysis of the Root Mean Square Error (RMSE). RMSE
is derive form the MSE it defines the actual ration of the
noise over the input signal or image. As RMSE needs to
be minimized; so the main attention is to reduce them
RMSE as much as possible. Table 6 and Figure 14 are
clearly shown that RMSE is less in our case therefore
proposed algorithm is providing quite effective results.

Table 6: RMSE
Sr.
No.
Image
Name
Old Proposed
1 A 9 6.4807
2 B 18.8944 11.9164
3 C 7.6811 7
4 E 8.1854 6
5 F 3.3166 2.8284
6 G 4.5826 3.1623
7 H 19.8746 11.8743
8 I 1.7321 1
9 J 5.6569 3.8730
10 K 11.4018 6.4807
11 L 1 0
12 M 12.6886 7
13 N 7.8740 4.6904
14 O 8.1240 4.8990
15 P 4.1231 2.2361
16 Q 2.6458 1.7321


Figure 14: RMSE Analysis
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com
Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 168


Table 7 and Figure 15 are showing the comparative
analysis of the Mean square error (MSE). As MSE need
to minimize; so our goal is to reduce them MSE as much
as possible. Table 7 and Figure 15 have clearly shown
that MSE is less in the case of the proposed algorithm;
therefore proposed algorithm is providing quite better
results.
Table 7: MSE analysis
Sr.
No.
Imag
e
Name
Old Proposed
1 a 81 42
2 b 357 142
3 c 59 49
4 e 67 36
5 f 11 8
6 g 21 10
7 h 395 141
8 i 3 1
9 j 32 15
10 k 130 42
11 l 1 0
12 m 161 49
13 n 62 22
14 o 66 24
15 p 17 5
16 q 7 3


Figure 15: Mean square error (MSE) evaluations

Table 8 and Figure 16 has shown the analysis of cross
correlation. It is required to be closed to 1. So the Table 8
and Figure 16 has clearly shown that the NCC is closed to
1 in case of the proposed algorithm. Therefore the
proposed algorithm has shown significant results over the
available methods.

Table 8: Normalized cross correlation (NCC)
Sr.
No.
Image name Old Proposed
1 a 0.9833 0.9994
2 b 0.9708 0.9872
3 c 0.9881 0.9928
4 e 0.9858 1.2135
5 f 0.9954 0.9994
6 g 0.9949 1.0068
7 h 0.9474 1.0052
8 i 0.9977 0.9790
9 j 0.9948 1.0081
10 k 0.9850 1.0108
11 l 0.9972 0.9985
12 m 0.9855 0.9985
13 n 0.9937 1.0124
14 o 0.9948 1.0043
15 p 0.956 0.9972
16 q 0.9956 1.0003


Figure 16: Normalized cross correlation (NCC)

9. CONCLUSION and FUTURE WORK
9.1 Conclusion
This paper has presented a modified image compression
algorithm using the Bayer layer and color normalization.
As decompression is done by using interpolation
techniques therefore may result in color artefacts and
corrupted pixels. The survey has shown that the
compression algorithm based on Bayer layer suffers from
color artefacts so it is required to reduce them. So this
work has proposed a modified algorithm with color
normalization. The proposed algorithm has been designed
and implemented in MATLAB using image processing
toolbox. The comparisons have shown that the proposed
algorithms provide quite significant results.

9.2 Future Work
In near future we will extend this work by integrating
colordemosaicing algorithms with different filtering
algorithms to enhance theresults. Also genetic
programming will also be considering interpolating the
missing colors in efficient manner.

References
[1] Jung, Miyoun, Xavier Bresson, Tony F. Chan, and
Luminita A. Vese. "Nonlocal Mumford-Shah
regularizers for color image restoration." Image
Processing, IEEE Transactions on 20, no. 6 (2011):
1583-1598.
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com
Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 169


[2] Popescu, Alin C., and HanyFarid. "Exposing digital
forgeries in color filter array interpolated images."
Signal Processing, IEEE Transactions on 53, no. 10
(2005): 3948-3959.
[3] Chung, King-Hong, and Yuk-Hee Chan. "A lossless
compression scheme for Bayer color filter array
images." Image Processing, IEEE Transactions on
17, no. 2 (2008): 134-144.
[4] Hirakawa, Keigo, and Patrick J. Wolfe. "Spatio-
spectral color filter array design for optimal image
recovery." Image Processing, IEEE Transactions on
17, no. 10 (2008): 1876-1890.
[5] Yasuma, Fumihito, TomooMitsunaga, Daisuke Iso,
and Shree K. Nayar. "Generalized assorted pixel
camera: postcapture control of resolution, dynamic
range, and spectrum." Image Processing, IEEE
Transactions on 19, no. 9 (2010): 2241-2253.
[6] Lee, Dongjae, Byung-JoonBaek, and Tae-Chan Kim.
"Two-layer color filter array for high quality images."
In Image Processing (ICIP), 2012 19th IEEE
International Conference on, pp. 345-348. IEEE,
2012.
[7] Hirakawa, Keigo, and Patrick J. Wolfe. "Spatio-
spectral color filter array design for optimal image
recovery." Image Processing, IEEE Transactions
on 17, no. 10 (2008): 1876-1890.
[8] Parmar, Manu, and Stanley J. Reeves. "Selection of
optimal spectral sensitivity functions for color filter
arrays." Image Processing, IEEE Transactions on 19,
no. 12 (2010): 3190-3203.
[9] Pekkucuksen,Ibrahim, and
YucelAltunbasak."Multiscale Gradients Based Color
Filter Array Interpolation." (2013):1-1.
[10] Lian, Nai-Xiang, Lanlan Chang, Yap-Peng Tan, and
VitaliZagorodnov. "Adaptive filtering for color filter
array demosaicking." Image Processing, IEEE
Transactions on 16, no. 10 (2007): 2515-2525.
[11] DIP [LAST VISITED] 29-JULY-2013 [ONLINE
AVAILABLE] https://encrypted-tbn0.gstatic.com.
[12] R. Lukac and K. N. Plataniotis, A normalized model
for color-ratio based demosaicking schemes, in Int.
Conf. on Image Process., 2004, vol. 3, pp. 1657
1660
[13] B. K. Gunturk, Y. Altunbasak, and R. M. Mersereau,
Color plane interpolation using alternating
projections, IEEE Trans. Image Process., vol. 11,
no. 9, pp. 9971013, Sep. 2002.
[14] J. E. Adams and J. F. Hamilton, Jr., Adaptive color
plan interpolation in single sensor color electronic
camera, U.S. Patent 5 506 619, Apr.9, 1996.
[15] X. Li, B. Gunturk, and L. Zhang, Image
demosaicing: A systematic survey, Proc. SPIEInt.
Soc. Opt. Eng., vol. 6822, p.-1- 15, Jan. 2008.
[16] R. Lukac and K. N. Plataniotis. Data adaptive filters
for demosaicking: a framework[J]. Consumer
Electronics, IEEE Transactions on. 2005, 51(2): 560-
570.
[17] P. Longere, X. Zhang, P. B. Delahunt and D. H.
Brainard. Perceptual assessment of
demosaicingalgorithm performance[J]. Proceedings
of the IEEE. 2002, 90(1): 123-132.
[18] B. E. Bayer, Color imaging array, U.S. Patent
3971065, July, 1976.
[19] J. Glotzbach, Y. Altunbasak, R. W. Schafer, and R.
M. Mersereau, Demosaicking: color filter array
interpolation, IEEE Signal Processing Magazine,
vol. 22, pp. 44- 54, Jan., 2005. 2007 0024879, Feb.
2007.

You might also like