Comprehensive Infrared Image Edge detection Algorithm

Abhishek Gudipalli & Ramashri Tirumala
International Journal of Image Processing (IJIP), Volume (6) : Issue (5) : 2012 297
Comprehensive Infrared Image Edge Detection Algorithm
Abhishek Gudipalli abhishek.g@vit.ac.in
Assistant Professor/School of Electrical Engineering
VIT University
Vellore,632014, India
Dr.Ramashri Tirumala rama.jaypee@gmail.com
Associate Professor/Dept of Electronics & Communication Engineering
Sri Venkateswara University College of Engineering
Tirupati,517501,India
Abstract
Edge detection is one of the most powerful image analysis tools for enhancing and detecting
edges. Indeed, identifying and localizing edges are a low level task in a variety of applications
such as 3-D reconstruction, shape recognition, image compression, enhancement, and
restoration. This paper introduces a new algorithm for detecting edges based on color space
models. In this RGB image is taken as an input image and transforming the RGB image to color
models such as YUV, YCbCr and XYZ. The edges have been detected for each component in
color models separately and compared with the original image of that particular model. In order to
measure the quality assessment between images, SSIM (Structural Similarity Index Method) and
VIF (Visual Information Fidelity) has been calculated. The results have shown that XYZ color
model is having high SSIM value and VIF value. In the previous papers, edge detection based on
RGB color model has low SSIM and VIF values. So by converting the images into different color
models shows a significant improvement in detection of edges.
Keywords: Edge detection, Color models, SSIM, VIF.
1. INTRODUCTION
The important image processing tool that is used as a fundamental pre-processing step in
many image processing applications is edge detection. Edges map have a significant role in
application such as image categorization, image registration, feature extraction, and pattern
recognition. An edge detector can be defined as a mathematical operator that responds to the
spatial change and discontinuities in gray levels of pixels set in an image. Each industry will use
its suitable color model. For example, the RGB color model is used in computer graphics, YUV is
used in video systems, PhotoCD production and so on. Edge detection in color images requires
several approaches of different complexity already exist. In image processing edge detection is
an important process, color images provides more detailed edge information than gray value
images, and color edge detection becomes vital for edge based image segmentation or edge-
based stereo matching. Edges will not be detected in gray value images when neighboring
objects have different hues but equal intensities. Additionally, the common shortcomings of the
RGB image edge detection arithmetic are the low speed and the color losses after the each
component of the image is processed. Thus, color edge detection is proposed based on changing
the domain of the edge detection. The RGB image is transformed to YUV, YCbCr and XYZ color
models. The quality assessment metrics such that SSIM and VIF have been applied to the above
models and found that in XYZ color model is providing more detailed edge information than the
other color models.

2. DIFFERENT COLOR SPACES
2.1 RGB color model
In the RGB model, each color appears as a combination of red, green, and blue. This model is
called additive, and the colors are called primary colors. The primary colors can be added to
produce the secondary colors of light (see Figure "Primary and Secondary Colors for RGB ") -
magenta (red plus blue), cyan (green plus blue), and yellow (red plus green). The combination of
red, green, and blue at full intensities makes white. The color subspace of interest is a cube
shown in Figure "RGB Color Model" (RGB values are normalized to 0..1)[1], in which RGB values
are at three corners; cyan, magenta, and yellow are the three other corners, black is at their
origin; and white is at the corner farthest from the origin. The gray scale extends from black to
white along the diagonal joining these two points. The colors are the points on or inside the cube,
defined by vectors extending from the origin. Thus, images in the RGB color model consist of
three independent image planes, one for each primary color. The importance of the RGB color
model is that it relates very closely to the way that the human eye perceives color. RGB is a basic
color model for computer graphics because color displays use red, green, and blue to create the
desired color. Therefore, the choice of the RGB color space simplifies the architecture and design
of the system. Besides, a system that is designed using the RGB color space can take advantage
of a large number of existing software routines, because this color space has been around for a
number of years. However, RGB is not very efficient when dealing with real-world images. To
generate any color within the RGB color cube, all three RGB components need to be of equal
pixel depth and display resolution. Also, any modification of the image requires modification of all
three planes[1].
FIGURE 1: RGB Colour model.
2.2 YCbCr and YUV color model
In the YCbCr color space is used for component digital video is a scaled and offset version of the
YUV color space. The YUV color model is the basic color model used in analogue color TV
broadcasting. Initially YUV is the re-coding of RGB for transmission efficiency (minimizing
bandwidth) and for downward compatibility with black-and white television. The YUV color space
is “derived” from the RGB space. It comprises the luminance (Y) and two color difference (U, V)
components. The luminance can be computed as a weighted sum of red, green and blue
components; the color difference, or chrominance, components are formed by subtracting
luminance from blue and from red. The principal advantage of the YUV model in image
processing is decoupling of luminance and color information. The importance of this decoupling is
that the luminance component of an image can be processed without affecting its color
component. For example, the histogram equalization of the color image in the YUV format may
be performed simply by applying histogram equalization to its Y component. There are many
combinations of YUV values from nominal ranges that result in invalid RGB values, because the
possible RGB colors occupy only part of the YUV space limited by these ranges. For example,

the histogram equalization of the color image in the YUV format may be performed simply by
applying histogram equalization to its Y component. There are many combinations of YUV values
from nominal ranges that result in invalid RGB values, because the possible RGB colors occupy
only part of the YUV space limited by these ranges. Figure " YUV Color Model" shows the valid
color block in the YUV space that corresponds to the RGB color cube RGB values are normalized
to [0..1])[1].
The Y’U’V’ notation means that the components are derived from gamma-corrected R’G’B’.
Weighted sum of these non-linear components forms a signal representative of luminance that is
called luma Y’. (Luma is often loosely referred to as luminance, so you need to be careful to
determine whether a particular author assigns a linear or non-linear interpretation to the term
luminance)[1].The YCbCr color space is used for component digital video is a scaled and offset
version of the YUV color space. The position of the block of RGB-representable colors in the
YCbCr space is shown in Figure2.1 RGB Colors Cube in the YCbCr Color Model [1].
Conversion between RGB and YCbCr models:
Y’ = 0.257*R' + 0.504*G' + 0.098*B' + 16
Cb' = -0.148*R' - 0.291*G' + 0.439*B' + 128
Cr' = 0.439*R' - 0.368*G' - 0.071*B' + 128
Conversion between RGB and YUV models:
Y'= 0.299*R' + 0.587*G' + 0.114*B'
U'= -0.147*R' - 0.289*G' + 0.436*B' = 0.492*(B'- Y')
V'= 0.615*R' - 0.515*G' - 0.100*B' = 0.877*(R'- Y')
FIGURE 2: YCbCr and YUV color models.
2.3 XYZ color model
The XYZ color space is an international standard developed by the CIE (Commission
Internationale de l’Eclairage). This model is based on three hypothetical primaries, XYZ, and all
visible colors can be represented by using only positive values of X, Y, and Z. The CIE XYZ
primaries are hypothetical because they do not correspond to any real light wavelengths. The Y
primary is intentionally defined to match closely to luminance, while X and Z primaries give color
information. The main advantage of the CIE XYZ space (and any color space based on it) is that
this space is completely device-independent. The position of the block of RGB-representable
colors in the XYZ space is shown in Figure 3 XYZ Color model [1].
Conversion between RGB and XYZ models:
X = 0.412453*R’ + 0.35758 *G’ + 0.180423*B’

Y = 0.212671*R’ + 0.71516 *G’ + 0.072169*B’
Z = 0.019334*R’ + 0.119193*G’ + 0.950227*B’
FIGURE 3: XYZ color model.
2.4 Infrared Imaging
Infrared (IR) light is electromagnetic radiation with a wavelength longer than that of visible light,
measured from the nominal edge of visible red light at 0.74µm, and extending conventionally to
300µm. These wavelengths correspond to a frequency range of approximately 1 to 400 THz and
include most of the thermal radiation emitted by objects near room temperature. Microscopically,
IR light is typically emitted or absorbed by molecules when they change their rotational and
vibration movements. Infrared imaging is used extensively for military and civilian purposes.
Military applications include target acquisition, surveillance, night vision, homing and tracking.
Non-military uses include thermal efficiency analysis, environmental monitoring, industrial facility
inspections, remote temperature sensing, short-ranged wireless communication, spectroscopy,
and weather forecasting[2]. Infrared astronomy uses sensor-equipped telescopes to penetrate
dusty regions of space, such as molecular clouds; detect objects such as planets, and to view
highly red-shifted objects from the early days of the universe. Humans at normal body
temperature radiate chiefly at wavelengths around 12 µm, at the atomic level infrared energy
elicits vibration modes in a molecule through a change in the dipole moment, making it a useful
frequency range for study of these energy states for molecules of the proper symmetry. Infrared
spectroscopy examines absorption and transmission of photons in the infrared energy range,
based on their frequency and intensity [2].
3. EVALUATION METRICS
In image processing applications, the measurement of image quality plays main role. Image
quality assessment algorithms are classified into three categories: FullReference (FR), Reduced-
Reference (RR), and No-Reference (NR) algorithms. True No Reference algorithms are difficult to
design and little progress has been made [3]. Full Reference algorithms are easier to design and
The SSIM index is a full reference metric. In this, the measurement of image quality is based on
reference image of perfect quality. SSIM is designed to improve Peak Signal-to-Noise Ratio
(PSNR) and Mean Squared Error (MSE), which is proved to be inconsistent with human eye
perception[4]. However, in RR or NR quality assessment, partial or no reference information is
available. The SSIM index is defined as[4]:
SSIMሺx,yሻ=
σxy+C1
σxσy+C1
.
2µx
µy
+ C2
µx
2+µy
2+ C2
.
2σxσy+ C3
σx
2
+σy
2
+ C3
Let x an y be the two discrete non-negative signals extracted from the same spatial location from
two images being compared, respectivelyߤ௫, ߪ௫
ଶ
and ߪ௫௬ be the mean of x ,the variance of x and
the covariance of x and y, respectively. ߤ௫ and ߪ௫ gives the information on luminance and
contrast of x. ߪ௫௬ measures the structural similarity.where C1, C2 and C3 are small constants given
by C1 = (K1 L)
2
; C2 = (K2 L)
2
and C3 = C2 / 2; respectively. L is the dynamic range of the pixel

values (L = 255 for 8 bits/pixel gray scale images), and K1 < 1 and K2 < 1 are two scalar
constants [4].
Sheikh and Bovik (2006) developed a visual information fidelity (VIF) index for Full Reference
measurement of quality of image. VIF is calculated between the reference image and its copy[5].
For ideal image, VIF is exactly unity. For distorted image types, VIF lies in between interval [0, 1].
Let e=c+n be the reference image, and n zero-mean normal distribution N (0, ߪ௡
ଶ
I ) noise. Also, let
f=d+n′= gc+v′+ n′ be the test image, where g represents the blur, v′ the additive zero-mean
Gaussian white noise with covariance ߪ௩
ଶ
I , and n′ the zeromean normal distribution N(0,σ n2I )
noise[3]. Then, VIF can be computed as the ratio of the mutual information between c and f, and
the mutual information between c and e for all wavelet subbands except the lowest approximation
subband[4].
VIF=
ΣI(c ;f|z)
ΣI(c ;e|z)
4. PROPOSED EDGE DETECTION ALGORITHM
The proposed algorithm can be explained in seven steps
Step 1: The RGB image is sub divided into R, G and B layers of the image.
Step 2: A 3X3 Laplacian mask is convolved with the R component of the image.
Step 3: The edge detected R and the G, B layers of the image are concatenated to obtain edge
detected image
Step 4: SSIM and VIF values are calculated between the R edge detected image and RGB
image.
Step 5: Repeat steps 2 to 4 to calculate the SSIM and VIF values between G edge detected
image and RGB image
Step 6: Repeat steps 2 to 4 to calculate the SSIM and VIF values between B edge detected
image and RGB image
Step 7: The SSIM and VIF values of individual components are averaged.
Step 8: R, G and B values of the image are transformed into its YCbCr, YUV and XYZ
Intensity values using the conversion formulas.
Step 9: Repeat steps 1 to 8 to calculate SSIM and VIF values for YCbCr, YUV and XYZ images.
5. EXPERIMENTAL RESULTS
The Proposed algorithm has been applied to RGB, YCbCr, YUV and XYZ images and SSIM &VIF
values are computed for a set of edge detected images and dataset is formed and tabulated in
Table 1. The property of infrared images is that intensity value depends on temperature, object
surface, surface direction, wavelength, etc. based up on these, for RGB color model, The SSIM
value range is around 0.57 and VIF value range is around 0.21. For YCbCr color model, The
SSIM value range is around 0.58 and VIF value range is around 0.13. For YUV color model, The
SSIM value range is around 0.49 and VIF value range is around 0.16. For XYZ color model, The
SSIM value range is around 0.61 and VIF value range is around 0.33.from the dataset, XYZ
model shows better quality of edge detection than the other color models. The original and edge
detected RGB, YCbCr, YUV and XYZ images are shown below.
Color model SSIM VIF
RGB 0.5730 0.2147
XYZ 0.6156 0.3306
YCbCr 0.5790 0.1316
YUV 0.4907 0.1675
TABLE 1: SSIM and VIF Values for RGB, XYZ, YCbCr and YUV images.

FIGURE 4: RGB Image
FIGURE 5: RGB edge detected image
FIGURE 6: XYZ image
Image
50 100 150 200 250 300 350 400 450 500
50
100
150
200
250
300
350
400
450
500
RGB image after Red component edge detection
XYZ image

FIGURE 7: XYZ edge detected image
FIGURE 8: YCbCr image
FIGURE 9: YCbCr edge detected image
XYZ image after X component edge detection
YCbCr image
YCbCr image after Y component edge detection

FIGURE 10: YUV image
FIGURE 11: YUV edge detected image
6. CONCLUSION
The proposed approach has a potential for various applications to detect edges of Infrared
images used extensively for military and civilian purposes. Humans at normal body temperature
radiate chiefly at wavelengths around 12 µm a comprehensive method can be developed for
pattern recognition models based on edge detection algorithms. The method shown is an new
approach to detect edges in different color space models. The algorithm was developed based on
RGB colour space and the significant features extracted by converting it into YUV, YCbCr and
XYZ models. Among these XYZ model shows better quality of edge detection.
7. REFERENCES
1. Image Color Conversion,
“http://software.intel.com/sites/products/documentation/hpc/ipp/ippi/ippi_ch6/ch6_Intro.html”.
2. Dong Wang and Jingzhou Zhang “Infrared image edge detection algorithm based on sobel
and ant colony algorithm”, 2011 International conference on multimedia technology, pp.4944
– 4947.
YUV image
YUV image after Y component edge detection

3. Sheikh, H. R., Bovik, A. C. and Cormack, L., “Noreference quality assessment using natural
scene statistics: JPEG2000”, IEEE Transactions on Image Processing, 14(11), 2005, pp.
1918-1927.
4. S Qian, G Chen “Four reduced-reference metrics for measuring hyperspectral images after
spatial resolution enhancement” In: Wagner W., Székely, B. (eds.): ISPRS TC VII
Symposium – 100 Years ISPRS, Vienna, Austria, July 5–7, 2010, IAPRS, Vol. XXXVIII, Part
7A
5. Sheikh, H. R. and Bovik, A. C, “Image information and visual quality”, IEEE Transactions on
Image Processing, 15(2), 2006, pp. 430-444.
6. Chen, G. Y. and Qian, S. E, “Evaluation and comparison of dimensionality reduction methods
and band selection”, Canadian Journal of Remote Sensing, 34(1), 2008a, pp. 26-32.
7. Li, Q. and Wang, Z., “Reduced-reference image quality assessment using divisive
normalization-based image representation”, IEEE Journal of Selected Topics in Signal
Processing, Special issue on Visual Media Quality Assessment, 3(2), 2009,pp. 202-211.
8. S K Naik , C A Murthy, “Standardization of edge magnitude in color images”, IEEE
Transactions on Image Processing , 2006 , 15(9) : 2588 —2595.
9. C L Novak, S A Shafer “Color edge detection”, Proc of DARPA Image Understanding
Workshop. 1987, pp:35 -37
10. R C Gonzalez,R E Woods, “Digital Image Processing” (Second Edition).New York : Prentice
Hall , 2003 pp.300-450.
11. P E Trahanias , A N Venetsanopoulos, “Color Edge Detection Using Vector Order Statistics”,
IEEE Transactions on Image Processing ,1993 , 2 (2) : 259 -264.
12. J Fan , W GAref , M Hacid , et al. “An improved automatic isotropic color edge detection
technique”, Pattern Recognition Letters,2001,22(3):1419-1429
13. A. Koschan, “A comparative study on color edge detection,” in Proceedings of the 2nd Asian
Conference on Computer Vision ACCV’95, pp. 574-578, Singapore, 1995.
14. A. N. Evans and X. U. Liu, “A morphological gradient approach to color edge detection,”
IEEE Transactions on Image Processing, vol. 15, no. 6, pp.1454-1463, 2006.
15. X. W. Li and X. R. Zhang, “A perceptual color edge detection algorithm,” International
Conference on Computer Science and Software Engineering, vol. 1, pp.297-300, 2008.
16. Yajun Fang, Keiichi Yamadac, Yoshiki Ninomiya, Berthold Horn, Ichiro Masaki “Comparison
between infrared image based and visible image based approaches for pedestrian detection”,
IEEE Intelligent Vehicles Symposium, pp.505-510,2003.
17. Kenji Omasa, Fumiki Hosoi and Atsumi Konishi,” 3D lidar imaging for detecting and
understanding Plant responses and canopy structure”, Journal of Experimental Botany,
Vol. 58, No. 4, pp. 881–898, 2007.

Comprehensive Infrared Image Edge detection Algorithm

More Related Content

What's hot (20)

Similar to Comprehensive Infrared Image Edge detection Algorithm (20)

Recently uploaded (20)

Comprehensive Infrared Image Edge detection Algorithm