Study On The Restoration Of Underwater Image

Underwater images are widely used in many fields such as water conservancy projects,river water quality environmental monitoring,and underwater robots.Due to the influence of light on the absorption and scattering of water,the quality of underwater images obtained is severely degraded.Therefore,it is an important research direction to use technical means to recover the underwater images and restore the original clear images to the maximum extent,thereby obtaining the required image information.The research work on underwater image restoration methods in this paper is as follows:The illumination received by the object under the water is not uniform,while the traditional Retinex algorithm assumes that the underwater object is uniformly illuminated,resulting in deviations in the illumination estimates of the underwater image,which in turn produces halos in the edge regions where the contrast between the light and shade is strong.Artifacts cause blurring of the edges of the image and loss of details.To solve these problems,this paper proposes to improve the filter function of the homomorphic filter algorithm,so that the improved homomorphic filter algorithm can process the high-frequency information and low-frequency information of the image separately.Thus,the area with strong image illumination is suppressed,the dark area of the image light is preserved,the brightness difference between the strong light area and the low light area is reduced,and the light unevenness of the image is corrected.Finally,the multi-scale Retinex algorithm is used for image deblurring to improve the overall restoration effect.The experimental results show that the proposed algorithm can effectively eliminate the halo blur of underwater images and enhance the image edges.It is verified that the recovery effect of the proposed algorithm is better than the traditional Retinex algorithm by calculating the evaluation indexes such as clarityUsing the dark channel prior method to recover the underwater image,most of the noise in the image can be removed.However,the output image has the characteristics of low brightness and contrast,and the edge blurring of the output image is not reduced.To address these deficiencies,this paper proposes an underwater image restoration algorithm based on dark channel prior and Clahe.Firstly,the guided filter algorithm is used to optimize the transmittance so that the computational complexity of the prior algorithm of the dark channel reaches O(N)(N pixels)while maintaining the edge smoothness of the image.Then the processed image is segmented,combined with the Clahe algorithm to enhance the brightness of the image,and further enhance the image edge,improving the clarity and contrast of the image.The experimental algorithm is compared with the common dark channel prior algorithm and the evaluation indexes such as running time and information entropy are calculated.The comparison results show that the algorithm presented in this paper is more real-time,and the sharpness and edge strength of the restored image are also higher.