I.                   
INTRODUCTION

Many important visual applications, such as
detection and scene understanding rely on the quality of the image. However it
is common that in bad weather conditions like haze , fog, rain image quality
degrades. Chromatic Atmospheric Scattering1
is a study done on different climatic conditions. The
traditional image processing techniques where not sufficient to remove weather
effects from images, thus they introduced a physics-based model that describes
the appearances of scenes in uniform bad weather conditions proposed a fast algorithm
2. Single image haze removal 3 uses the
Gaussian -based method because the original image has very low intensities. Physical model4 is used in a computer vision in
order to have a haze free image in different climatic conditions earlier they
used polarization now to have a better enhancement in the image a physical
model is used. The image defogging 5 is occurred
due to frequently exposure to strong light, rain, snow and fog thus
Gaussian-based dark channel is proposed.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

II.               
LITERATURE REVIEW

Shree K. Nayar and Srinivasa G. Narasimhan in the year 1999
explains all about the bad weather and they have studied different weather
conditions and identifying effects caused by the poor weather and trying to
make the effects as advantages and the images where transformed into the three
dimensional structure. 1

Srinivasa G. Narasimhan and Shree K. Nayar in the year 2003
explained about the right amount of light required to the camera and they have
proposed technical method to have a good image. In technical method they used
fast algorithm to overcome the pollutants in the images. 2

 

Kaiming He, Jian Sun, Xiaoou Tang in the year 2011 tell us
about prior-dark channel, to remove all the pollutants such as haze, rain,
rain, snow from a single input image. They use a high-quality haze free image
to remove all the pollutants. 3

 

Qingsong Zhu, Jiaming Mai, Ling Shao in the year 2015
address about the color attenuation prior in this paper to remove the haze from
a single input image , they use depth map and restore the radiance through the
atmospheric light. 4

 

Jing-Ming, Guo, Jin-yuSyue, VincentRadzicki, HuaLee, Fellow
in the year 2017 wrote a paper based on defogging , it deals with different
algorithm Li, Tarel, Hazy, He, Meng, Lai, Zhu, Tang , Kim , Kolor , Proposed
method, Gaussian based dark channel , the problem in the image was caused by
defogging to overcome this problem they used Fusionbased transmission
estimation method combined with two different transmission models , The new
fusion weighting scheme, the atmospheric light computed from the Gaussian-based
dark channel and the flicker-free module. To get a complete defog image, the
dehazing was combined with fusion weighting function and Vibe method was used
to remove or reduce the flicker effect in the image. 5

 

 

 

 

 

III.              
PROPOSED METHOD

 

The algorithm for restoring hazy
images in various levels is shown in Figure 1.We first load the image which is
captured in the outdoor scene, then we separate the RGB components and compute
the Gaussian -based
dark channel prior, with the hazy image and assumption of atmospheric light we
can Estimate the transmission. To enhance the image Laplance Transformation is estimated,
thus haze free image is restored.

 

               Fig.1Proposed Method Taxonomy

 

 

 

 

3.1  RGB
COMPONENT

 

The RGB components are separated in
three colours red, green and blue. Here the original image is split into three
images called red, green and blue image

 

3.2  Gaussian-based
dark channel prior

 

The Gaussian-based dark channel is
proposed in the given original image to identify the RGB colour component in
each pixel of the original image. 

 

 

3.3  Estimated
Transmission

 

The Estimated Transmission is called
A is proposed in the original image to estimate the atmospheric light.

    

3.4  Laplance
Transformation

 

Laplacian filter is used to remove
disturbance in the original image and it highlights the edges of the image.

 

 

IV.            
PERFORMANCE ANALYSIS

 

In
this paper we
have proposed Haze image using Gaussian- based dark channel algorithm. The
Gaussian-based dark channel is outdoor haze based images, which contain dark
pixels due to bad weather. Our proposed method consist of RGB component, The
dark channel and estimated transmission which is a atmospheric light and
Laplance filter to enhance the haze image. The image pixel is visualized using
image tool for each output image to analysis the pixel count is the image. The
image tool shows the pixel count of each RGB component present in the image.
PSNR(Peak signal to Noise Ratio),The PSNR is used to calculate the maximum
pixel values in the original image and the proposed method. The PSNR value
should be maximum in the proposed image compared to the original image.
MSE(Mean Square Error), The MSE is used to calculate the error in the original
image and the proposed image. The MSE value should be minimum in the proposed
image compared to the original image.

      Fig.2.
Results of performance analysis

 

 

 

V.               
CONCLUSION

To establish a dehaze model, the RGB and
Gaussian-based dark channel method was proposed. The further method consist of
estimated transmission and Laplance Transformation to dehaze the image which is
taken in the poor weather conditions. The PSNR and MSE is calculated in the
proposed method to know the performance in the original image and the proposed
image.

 

REFERENCES:

 

1 S. K. Nayar and S. G. Narasimhan, "Vision in
BadWeather," in Proc. IEEE Int. Conf. Comput. Vis.(ICCV), vol. 2. Sep.
1999, pp. 820–827.

 

2 S. Narasimhan and S. Nayar, " Contrast
Restoration of Weather Degraded Images", IEEE Transactions on Pattern
Analysis and Machine Intelligence, vol. 25, no. 6, pp.713-724, 2003.

 

3 K. He, J. Sun, and X. Tang, “Single Image Haze
Removal Using Dark Channel Prior,” IEEE
Trans. Pattern Anal. Mach.Intell., vol. 33, no. 12, pp. 2341–2353, Dec. 2011.

 

 

4 Q. Zhu, J. Mai and L. Shao, “A Fast Single Image
Haze Removal Algorithm Using Color Attenuation Prior,” Image Processing, IEEE Transactions on, on
page(s): 3522 – 3533 Volume: 24, Issue: 11, Nov. 2015.

 

5 Jing-Ming Guo, Senior Member, IEEE, Jin-yuSyue,*Vincent
Radzicki, and *Hua Lee, Fellow,” An EfficientFusion-Based Defogging”, IEEE IEEE
Transactions on ImageProcessing ,Vol: 26, Issue: 9, Sept. 2017

 

Written by
admin
x

Hi!
I'm Colleen!

Would you like to get a custom essay? How about receiving a customized one?

Check it out