I. INTRODUCTIONMany important visual applications, such asdetection and scene understanding rely on the quality of the image. However itis common that in bad weather conditions like haze , fog, rain image qualitydegrades.
Chromatic Atmospheric Scattering1is a study done on different climatic conditions. Thetraditional image processing techniques where not sufficient to remove weathereffects from images, thus they introduced a physics-based model that describesthe appearances of scenes in uniform bad weather conditions proposed a fast algorithm2. Single image haze removal 3 uses theGaussian -based method because the original image has very low intensities. Physical model4 is used in a computer vision inorder to have a haze free image in different climatic conditions earlier theyused polarization now to have a better enhancement in the image a physicalmodel is used. The image defogging 5 is occurreddue to frequently exposure to strong light, rain, snow and fog thusGaussian-based dark channel is proposed.II. LITERATURE REVIEWShree K.
Nayar and Srinivasa G. Narasimhan in the year 1999explains all about the bad weather and they have studied different weatherconditions and identifying effects caused by the poor weather and trying tomake the effects as advantages and the images where transformed into the threedimensional structure. 1Srinivasa G. Narasimhan and Shree K. Nayar in the year 2003explained about the right amount of light required to the camera and they haveproposed technical method to have a good image.
In technical method they usedfast algorithm to overcome the pollutants in the images. 2 Kaiming He, Jian Sun, Xiaoou Tang in the year 2011 tell usabout 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 imageto remove all the pollutants. 3 Qingsong Zhu, Jiaming Mai, Ling Shao in the year 2015address about the color attenuation prior in this paper to remove the haze froma single input image , they use depth map and restore the radiance through theatmospheric light. 4 Jing-Ming, Guo, Jin-yuSyue, VincentRadzicki, HuaLee, Fellowin the year 2017 wrote a paper based on defogging , it deals with differentalgorithm Li, Tarel, Hazy, He, Meng, Lai, Zhu, Tang , Kim , Kolor , Proposedmethod, Gaussian based dark channel , the problem in the image was caused bydefogging to overcome this problem they used Fusionbased transmissionestimation method combined with two different transmission models , The newfusion weighting scheme, the atmospheric light computed from the Gaussian-baseddark channel and the flicker-free module. To get a complete defog image, thedehazing was combined with fusion weighting function and Vibe method was usedto remove or reduce the flicker effect in the image.
5 III. PROPOSED METHOD The algorithm for restoring hazyimages in various levels is shown in Figure 1.We first load the image which iscaptured in the outdoor scene, then we separate the RGB components and computethe Gaussian -baseddark channel prior, with the hazy image and assumption of atmospheric light wecan Estimate the transmission. To enhance the image Laplance Transformation is estimated,thus haze free image is restored. Fig.
1Proposed Method Taxonomy 3.1 RGBCOMPONENT The RGB components are separated inthree colours red, green and blue. Here the original image is split into threeimages called red, green and blue image 3.2 Gaussian-baseddark channel prior The Gaussian-based dark channel isproposed in the given original image to identify the RGB colour component ineach pixel of the original image. 3.3 EstimatedTransmission The Estimated Transmission is calledA is proposed in the original image to estimate the atmospheric light. 3.
4 LaplanceTransformation Laplacian filter is used to removedisturbance in the original image and it highlights the edges of the image. IV. PERFORMANCE ANALYSIS Inthis paper wehave proposed Haze image using Gaussian- based dark channel algorithm. TheGaussian-based dark channel is outdoor haze based images, which contain darkpixels due to bad weather. Our proposed method consist of RGB component, Thedark channel and estimated transmission which is a atmospheric light andLaplance filter to enhance the haze image.
The image pixel is visualized usingimage tool for each output image to analysis the pixel count is the image. Theimage 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 maximumpixel values in the original image and the proposed method. The PSNR valueshould 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 originalimage and the proposed image. The MSE value should be minimum in the proposedimage compared to the original image.
Fig.2.Results of performance analysis V. CONCLUSIONTo establish a dehaze model, the RGB andGaussian-based dark channel method was proposed.
The further method consist ofestimated transmission and Laplance Transformation to dehaze the image which istaken in the poor weather conditions. The PSNR and MSE is calculated in theproposed method to know the performance in the original image and the proposedimage. REFERENCES: 1 S. K. Nayar and S.
G. Narasimhan, "Vision inBadWeather," in Proc. IEEE Int. Conf. Comput. Vis.
(ICCV), vol. 2. Sep.
1999, pp. 820–827. 2 S. Narasimhan and S. Nayar, " ContrastRestoration of Weather Degraded Images", IEEE Transactions on PatternAnalysis and Machine Intelligence, vol. 25, no. 6, pp.713-724, 2003.
3 K. He, J. Sun, and X.
Tang, “Single Image HazeRemoval Using Dark Channel Prior,” IEEETrans. Pattern Anal. Mach.Intell.
, vol. 33, no. 12, pp. 2341–2353, Dec. 2011.
4 Q. Zhu, J. Mai and L. Shao, “A Fast Single ImageHaze Removal Algorithm Using Color Attenuation Prior,” Image Processing, IEEE Transactions on, onpage(s): 3522 – 3533 Volume: 24, Issue: 11, Nov. 2015. 5 Jing-Ming Guo, Senior Member, IEEE, Jin-yuSyue,*VincentRadzicki, and *Hua Lee, Fellow,” An EfficientFusion-Based Defogging”, IEEE IEEETransactions on ImageProcessing ,Vol: 26, Issue: 9, Sept. 2017