Research On Underwater Image Clarification Algorithm Combining Saliency Information

Underwater image is an important source for humans to obtain marine information.When light travels in water,due to the selective absorption of light by water and scattering of light by particles in water,underwater images usually have defects such as color distortion blurred details,low brightness and low contrast,which seriously affect the application of underwater images in military scientific research and engineering projects.Therefore,the research on the clarification processing of underwater images is of great significance.In response to the above research tasks,this thesis mainly focuses on the following two tasks.Aiming at the problem of color distortion and low contrast of underwater images,an underwater image clearing algorithm that combines CIELab color information and multi-layer cellular automata(Multi-Layer Cellular Automata,MCA)saliency information is proposed.On the basis of combining the underwater imaging model for the preliminary image clarity,the fusion of multi-scale superpixel segmentation and multi-layer cellular automata is used to detect the saliency of underwater images Finally,the color correction of the underwater image is performed according to the saliency information.This algorithm can effectively alleviate the color distortion of underwater images and enhance the local contrast of the underwater image.Aiming at the problem of color distortion and low light in underwater images,an underwater image clarity algorithm is proposed that integrates retinal cortex theory and saliency information of boundary connectivity.Based on the superpixel segmentation of underwater images,the algorithm combines boundary connectivity,background weight contrast and smoothness,and applies the least square method to obtain a globally optimized saliency map.According to the obtained saliency information,a variety of enhancement algorithms based on the theory of retinal cortex are proposed.And the brightness value of the underwater image is stretched,and finally a clear underwater image is obtained.The algorithm can effectively improve the brightness,restore detail information,and perform color correction and contrast enhancement of the low-illumination underwater images.This thesis selects 1500 underwater images in the UFO-120 data set for experiments,and compares the above two algorithms with six representative underwater image sharpening algorithms and a classic low-light enhancement algorithm.Experimental results show that the two algorithms proposed in this thesis have obvious advantages in color correction,contrast and brightness enhancement.