Night image enhancement using fusion technique

IJRET: International Journal of Research in Engineering and Technology ISSN: 2319-1163
__________________________________________________________________________________________
Volume: 02 Issue: 05 | May-2013, Available @ http://www.ijret.org 807
NIGHT IMAGE ENHANCEMENT USING FUSION TECHNIQUE
Amritpal Singh1
, Vijay Kumar Banga2
1
Amritpal Singh, 2
Dr. Vijay Kumar Banga, E.C.E Department, A.C. E.T, Amritsar, Punjab, India
apspunn@gmail.com, vijaykumar.banga@gmail.com
Abstract
Video surveillance is used in monitoring the road traffic. However because the surveillance system is limited by many objective
factors. Surveillance video may not be seen clearly. Especially under the weak light conditions, the picture quality of the night video is
very poor.
In this paper, we propose an enhancement method of nighttime images for surveillance camera. We apply the moving target extraction
technology and illumination estimation theory and combine the nighttime image with the daytime background by the image fusion.
The proposed method recoveries scene information of the night video and highlights the details. The resultant images show the
vehicles of the night time but surrounding is fused from the day image taken at day. In this, it is observed that the PSNR value is 65.9
dB is very high and MSE is 0.128 which is very low as compared with earlier reported night image enhancement technique [1].
Experimental results prove that our method is effective.
Index Terms: Image enhancement, Motion detection, Image fusion
-----------------------------------------------------------------------***-----------------------------------------------------------------------
1. INTRODUCTION
Night video enhancement is one of the most important and
difficult component of the video security surveillance system.
Most images taken from scenes with non uniform distributed
illumination show the problem of being too contrasty. The
images then can be divided into several different regions
according to their need for enhancement [2]. Until recently a
grey- or green scale representation of night vision imagery has
been the standard. However, the increasing availability of
fused and multiband infrared and visual night vision systems
has led to a growing interest in the color display of night
vision imagery [3, 4, 5, 6 and 7]. People have difficulty in
understanding nighttime video because of the following
reasons: Firstly, due to reasons of sensor noises or low
luminance, night images appear much noise. Secondly, the
brightness distributes unevenly because of the factitious
illumination at night. In view of the above problems, the
research on the video enhancement technology at night is
meaningful. Nowadays, many techniques for image
enhancement are discussed, such as contrast stretching,
slicing, histogram equalization etc. The enhancement results
of the traditional algorithm are not ideal for the nighttime
video; these algorithms may cause excessive exposure or
amplify noise. Therefore, enhancement algorithms combined
with daytime image have attracted attention of many
researchers [8, 9, 10].
In this paper, we propose an enhancement method of
nighttime images for a surveillance camera. We not only apply
the moving target extraction technology, Retinex theory and
fusion method, but also add an index mapping function to the
fusion image. The objective of our method guarantees that
most of the important contexts in the scene are synthesized to
create a much clearer video for observers.
We present a novel technique for night image enhancement
technique which overcomes very low intensity images to a
good intensity image, in order to give better visibility to the
inspectors inspecting the highways. This night image
enhancement technique is based on hybrid model of day and
night images for the same area of that highway which is under
surveillance. This technique applies the hybrid image to
increase the contrast of local part of the image which is stable
(not moving objects) in the night images to that of non-stable
(moving objects) on the highway [11].
2. PROPOSED METHOD
The objective of the proposed method is to increase the
intensity of the night time images and to provide for
information about the captured area and increase the
situational awareness. The proposed method is
computationally fast and simple.

__________________________________________________________________________________________
Figure 1: Flow chart of methodology
2.1 ALGORITHM
First step: Initialize the camera and Frame/image captured
from digital camera with fps more than 18.
Second step: Hybrid (Day time & Night time) Technique
For using the hybrid model for night image enhancement
technique, it is important to capture at least five hundred
images of local area at day time (i.e. high intensity image) and
same for the night time (i.e. low intensity image).
Here we have variables with name
P= Number of pixels of an image (size of every image),
Num= number of images in data base
DTM=day time matrix (4D matrix to store images in one
matrix)
Storing the day time images in one single matrix
Now after storing the images for day time in one matrix DTM.
We have to take a mean image so that non-moving objects (i.e.
stationary objects) extracted in one image DM.
Storing the day time images in one single matrix NTM=Night
time matrix (4D matrix to store images in one matrix)
Now after storing the images for night time in one matrix
NTM. We have to take a mean image so that moving objects
(non-stationary objects) extracted in one image NM.
Taking live images with the variable name img where n is the
nth number of image form the starting and subtracting from
the night mean image to extract non-stable objects in night.
Removing the noise from DIFF image below 9 threshold
value. Storing the dimensions as row and Column of image in
Row and Column variables and replacing rest all with day
time mean image (DM).
In the next part, have done the fusion of both images i.e. high
intensity image with low intensity image to get the
enhancement of image.
Third step: Calculation of night image enhancement
algorithm with threshold variable „α‟.
3. RESULTS
In our work of night image enhancement, we take the image of
the size 640x480. We consider two metrics to show the
experimental results of the work. Metrics are PSNR and MSE.
The graphs show that the resultant enhanced image has
increased PSNR and decreased MSE as compared to the
original night image.

__________________________________________________________________________________________
0 10 20 30 40 50 60 70 80 90 100
0
1
2
3
mse
0 10 20 30 40 50 60 70 80 90 100
40
50
60
70
psnr
The graphs showing PSNR and MSE of original night image
and enhanced image has been given in this paper for 100
frames. The resultant enhanced image for given night image
has also been shown.
Figure 2: PSNR and MSE after applying fusion technique
Figure 3. Day Time Image
Figure 4. Night Time Image
Figure5: Enhanced Image Using Fusion Technique
Figure 3 represents that image is captured at day time. In this
image two corks are used as stationary objects.
Figure 4 represents that image is captured at night time. In this
image two corks are used as stationary objects and car is as
moving object.
Figure 5 represents that image is enhanced with the use of
fusion technique. In this image, the car is taken from night
image and background image which have two corks as
stationary objects taken from day time image. Both of these
images are fused to get the enhanced image.
CONCLUSION
Night image enhancement techniques are widely applicable to
different fields. The proposed method has given good results
in terms of MSE and PSNR. This algorithm can be used even
in darkness. By changing the threshold values different views
and different objects can be seen for same image by the users.
The noise has also been properly removed from the images.
The resultant enhanced image closely resembles the day image
by using this algorithm.
ACKNOWLEDGEMENTS
The authors are thankful to Dhiraj Kumar Singh and
Gagandeep Singh, Assistant Professor, Rayat Institute of
Engineering & Information Technology, Railmajra for
providing technical supports during preparation of the
manuscript.
REFERENCES
[[1] Vinay G. Vaidya, Chandrashekhar N. Padole “Night
Vision Enhancement Using Wigner Distribution”, IEEE 2008.
[2] Xiaoying Fang, Jingao Liu, Wenquan Gu, Yiwen Tang “A
Method to Improve the Image Enhancement Result based on
Image Fusion”,IEEE 2011.
[3] Li, G. and Wang, K., Applying daytime colours to
nighttime imagery with an efficient colour transfer method, In:
J.G. Verly & J.1. Guell (Ed.), Enhanced and Synthetic Vision
stationary parts

__________________________________________________________________________________________
2007, pp. 65590L-1-65590L-12, The International Society for
Optical Engineering, Bellingham, MA, 2007.
[4] Shi, J., Jin, W., Wang, L. and Chen, H., Objective
evaluation of colour fusion of visual and IR imagery by
measuring image contrast, In: H. Gong, Y. Cai & J.-P. Chatard
(Ed.), Infrared Components and Their Applications, pp. 594-
601, The International Society for Optical Engineering,
Bellingham, MA, 2005.
[5] Shi, 1.-S., Jin, W.-Q. and Wang, L.-X., Study on
perceptual evaluation of fused image quality for colour night
vision, Journal ofInfrared and Millimeter Waves, 24(3) ,pp.
236-240,2005.
[6] Tsagaris, V. and Anastasopoulos, D., Multispectral image
fusion for improved RGB representation based on perceptual
attributes, International Journal of Remote Sensing, 26(15)
,pp. 3241-3254, 2006.
[7] Zheng, Y., Hansen, B.C., Haun, A.M. and Essock, E.A.,
Colouring night-vision imagery with statistical properties of
natural colours by using image segmentation and histogram
matching, In: R. Eschbach & G.G. Marcu (Ed.), Colour
imaging X: processing, hardcopy and applications, pp. 107-
117, The International Society for Optical Engineering,
Bellingham, WA, 2005.
[8] Ramesh Raskar,Adrian llie and Jingyi Yu, "Image Fusion
for Context Enhancement and video surrealism" ,The 3rd
International Symposium on Non-Photorealistic Animation
and Rendering (NPAR), Annecy, France, 2004.
[9] I.Li, S. Z.Li, Q. Pan, and T. Yang "1l1umination and
Motion-Based Video Enhancement for Night Surveillance” In
Proc. of the 2nd Joint IEEE International Workshop on YS-
PETS,pages 169-175. Beijing, China, October 2005.
[10] Akito Yamasaki, Hidenori Takauji, Shun' ichi Kaneko,
"Denighting: Enhancement of Nighttime Images for a
Surveillance Camera", Kita 14, Nishi 9, Kita-ku, Sapporo 060-
0814, JAPAN.
[11] Manpreet Kaur, Sukhwinder Singh “Night Image
Enhancement using Hybrid of Good and Poor Images” IJCTA,
July-August 2012.
BIOGRAPHIES:
Amritpal Singh received the B.Tech degree
in Electronics and Communication
Engineering from Punjab Technical
University, Jallandhar. Presently, he is
working as Lecturer in RIEIT, Railmajra as
well as pursuing M.Tech in ECE from
ACET, Amritsar. His current research
interests include Image Processing.
Dr. Vijay Kumar Banga received the
B.Tech degree in Electronics and
Communication engineering from Punjabi
University Patiala, M.Tech degree in
Electronics and Communication
Engineering from Panjab University
Chandigarh and Phd. From Thapar
University Patiala. He is currently working as HOD in
Amrisar College of Engineering and Technology, Amritsar.
His current research interests include Soft Computing, Image
Processing, Autonomous Robots.

Night image enhancement using fusion technique

More Related Content

What's hot (17)

Viewers also liked (19)

Similar to Night image enhancement using fusion technique (20)

More from eSAT Journals (20)

Recently uploaded (20)

Night image enhancement using fusion technique