Research On Underwater Image Correction Technology

The underwater imaging technology,as one of the important means for human beings to explore the ocean world,has been widely used in Marine geological exploration,underwater target identification and Marine military and other fields.However,the random fluctuation of the water surface and the complexity of the underwater environment will cause problems such as geometric distortion,color shift,and low contrast in the observed image.Therefore,it is of great significance to carry out correction research on underwater images.At present,the research methods of enhancement correction for underwater image with low contrast and color distortion are mainly divided into two types: physical model and nonphysical model.Based on physical and non-physical models.The method of image enhancement based on physical model is based on the principle of underwater optical imaging,blind estimation of illumination intensity and transmittance image,and then solving the restoration image with high contrast and color balance.The image enhancement method based on the non-physical model mainly operates directly on the pixels of the degraded image and improves the image quality by global adjustment of the gray value.For geometrically distorted underwater images,the correction method is iterative registration of the image sequence,using the mean image as the reference image,and after multiple cyclic interpolation processes,the problem of image distortion and distortion can be effectively solved.This paper has conducted related research on the above-mentioned underwater image degradation and distortion,and the main work is as follows:(1)A multi-scale underwater image enhancement scheme based on illumination components is designed.This method is based on Retinex theory and underwater optical imaging model,which approximates the illumination component equivalent to the transmittance image.Combined with the principle of image composition,a proportional fusion method is proposed to make the transmittance image closer to the actual.Finally,the Gaussian pyramid is used.Multi-scale optimization of transmittance images.(2)An underwater image enhancement scheme based on underwater color image quality assessment(UCIQE)function is designed.This method uses the quadtree search algorithm to estimate the ambient light intensity;the assumption is that the local area transmittance is constant,and the image is divided into blocks.Within each image block,an objective function with a transmittance variable is constructed by UCIQE and the information loss function,And optimal solution;use guide filtering to clear the edges of the transmittance image.Experimental results show that the above two schemes can effectively solve the problem of color shift and low contrast of degraded underwater images.(3)The iterative image registration technique with pixel as the feature space is studied.Firstly,its main components are introduced in detail: interpolation function,measurement function and optimization function;then the processing flow of the algorithm is analyzed in depth;finally,an underwater distortion video correction software is independently designed and developed.After experimental testing,the software can better solve the problem of image geometric distortion and achieve the predetermined effect.