Underwater Image Enhancement Based On Color Correction And Dark Channel Prior

When propagating under the water,the attenuation of light with different wavelengths is different because of absorption.The decay of red light is the fastest,followed by green light,and blue light decays the most slowly,resulting in serious color distortion for underwater images with green-blue tone.In addition,the scattering effect of particles in water makes underwater images with low contrast and fuzzy.Underwater image enhancement technology aims to solve the above defects through a certain enhancement processing to get a distinct image with natural color and high contrast,which is good for human perception and computer analysis.At present,underwater image enhancement technology has been widely used in marine environmental monitoring,marine resources exploration and military combat,etc.In the paper,through extensive research of references at home and abroad,the principle of underwater optical imaging is firstly presented,and the difficulties for the underwater image enhancement technology are analyzed.Then,combined with various areas of digital image processing technology,the underwater image enhancement method is realized with color correction and dark channel prior.The main work of the paper is as follows:(1)The background and significance of underwater image enhancement is introduced in detail.Research approach and status at home and abroad of underwater image enhancement are analyzed and presented with the summary of shortcomings between some existing algorithms.The theory of underwater image enhancement is described:firstly,based on the underwater optical imaging model,through analysis of the characteristics of underwater images,the key to the underwater image enhancement technique is deduced----how to effectively correct the color distortion and de-blur the image.Then several classic underwater image enhancement algorithms are introduced with brief description of implementation process,and comparison of the advantages and disadvantages between different algorithms from theoretical viewpoint.(2)The dark channel priori theory and the image de-haze algorithm based on the theory are described and realized with brief exhibition of its enhancement effect by the processing results.Then,according to the similarity between the underwater optical imaging characteristics and the atmospheric light scattering model,the feasibility of applying the dark channel prior theory to the underwater image enhancement is analyzed.Through the corresponding simulation experiment,it can be found that how to effectively correct the color distortion of the underwater image is the key to the effect of applying the theory to underwater image enhancement method.(3)Research status of underwater image color correction and related theories are briefly introduced.Several existing color correction algorithms are given with brief summary of implementation process and the relative results.The color correction and dark channel priori theory provide the theoretical basis for the proposed algorithm.(4)The realization of proposed underwater image enhancement algorithm based on color correction and dark channel priori theory is presented.In this paper,a white balance color correction algorithm is proposed based on the assumption of gray world hypothesis and perfect reflection,which is a preprocessing step to correct the color distortion of underwater images.In order to make the dark channel priori theory more suitable for underwater image enhancement,the relevant steps of the algorithm are improved.Firstly,the guided filtering instead of the original soft-maturing algorithm is used to refine the estimated transmittance;secondly,quad-tree hierarchical search and dark channel theory are combined to improve the accuracy of the estimated background light intensity.Finally,Rayleigh stretch is used for the initially enhanced underwater image to improve the contrast before the final output of the image.(5)A number of underwater images are selected for the corresponding simulation experiments,and the proposed algorithm is compared with several existing algorithms from subjective and objective indicators.The experimental results show that compared with traditional algorithms,the proposed method can achieve better visual effect and higher objective indexes,which is an effective underwater image enhancement algorithm.