Research Of Underwater Color Image Enhancement Method

The degradation of image is the main limitation for the application of the wharf underwater structure visual inspection technique.The underwater structure of the wharf,due to the longterm impact of wave washing and large equipment loading,regularly appears different degrees of appearance defects,which seriously affects the production safety of the wharf.The inspection of defects in the appearance of underwater structures is usually carried out by means of manual probing by divers,which has the problems of non-objective detection results and high danger.The utilization of visual inspection technology is a method being explored by terminal operators in recent years,but since the underwater areas of terminals are generally turbid,the visual quality of the collected underwater images is low,and the size information of the defects in the appearance of underwater structures at terminals cannot be accurately obtained,so the key issue of using computer vision technology for the inspection of the appearance of underwater structures at terminals is how to effectively improve the visual quality of underwater.Therefore,this thesis promotes the practical application of visual measurement technology in the inspection of the appearance of underwater structures at terminals by studying effective underwater image enhancement methods to improve the visual quality of underwater structure appearance images.The main research of this thesis is summarized as follows:(1)In order to decide the dominant color-channel,compensated color-channel,and compensating coefficient accordingly,considering the color balance fusion enhancement method proposed by Ancuti et al.,and the “Two-Step” strategy proposed by Xueyang Fu et al.,this thesis proposes a “New Two-Step”(NTS)underwater image enhancement method,which contains an evaluation index preposition based channel compensation and white-balance color correction algorithm and an artificial multi-underexposure fusion detail reconstruction algorithm.First,the NTS sets different compensating approaches and coefficients,utilizing the global color compensating approach to get color balanced images respectively.Second,the NTS utilizes a comprehensive evaluation algorithm based on two objective visual quality metrics,to get the color corrected image with the highest score.Then,the NTS uses gamma adjustment to get five versions with under-exposure.Meanwhile,the NTS also utilizes Contrast Limited Adaptive Histogram Equalization(CLAHE)to process the version with gamma equaling to 1.Last,the NTS sets “Contrast”,“Saturation”,and “Exposure” as three weights,being blended with 5 versions into the multi-scale fusing scheme,to get the final enhanced result.(2)In order to enhance wharf underwater structure appearance images with extremly color distortion,this thesis proposes an “Adaptive Color Adjustment”(ACA)underwater image enhancement method.First,the ACA utilizes a color channel adaptive reduction algorithm to adjust the pixel intensities of the input.Second,the ACA also uses the comprehensive evaluation algorithm to correct the color information of the adjusted image.Meanwhile,the ACA stretches the intensities of the uncompensated color-channels.Then,the ACA also utilizes gamma adjustment operation to get 6 versions with different degrees of exposure.Last,the ACA only sets “Contrast” and “Exposure” as two weights,being blended with 6 versions into the multi-scale fusing scheme,to get the final enhanced result.(3)Wharf underwater structure appearance images with extremely color distortion and blurriness,enhanced by the ACA appear hazy probably.To address these issues,this thesis also proposes an underwater image enhancement method based on Guided Filtering Separated Computation(GFSC).First,GFSC utilizes the parameter adaptive adjustment based color correction algorithm,to get the color-corrected version of input image.Meanwhile,GFSC also uses the entropy adaptive adjustment based artificial multi-exposure fusion algorithm,to generate a contrast-enhanced version of the input image.Sencond,the GFSC employs the guided filter separation algorithm to generate a smooth base layer and an edge information containing detail layer.Then,GFSC just uses the Color Channel Transfer(CCT)algorithm to transfer color information from the color-corrected version to the base layer.Finally,GFSC just uses the image reconstruction algorithm to get the final enhance result.(4)This thesis utilizes the proposed method and 7 famous underwater image deblurring methods proposed in recent 5 years,to make a deblurring process on the wharf underwater structure appearance monocular and binocular images.Then,this thesis makes the comparative trial of the image visual quality for the processed wharf structure appearance monocular images,also the comparative experiment of the results of key-points extraction and matching for the processed wharf structure appearance binocular images.The related experimental results show that the underwater image enhancement proposed by this thesis could more effectively enhance the wharf underwater structure appearance monocular images.The proposed method could also improve the results of key-points extraction and matching for wharf underwater structure appearance binocular images more effectively.These could effectively promote the the practical application of visual measurement technology in the appearance inspection of underwater structures of wharves.