| As the country’s requirements for the development and exploration of marine resources have gradually increased in recent years,underwater vision technology has also developed rapidly.Among them,the underwater 3D reconstruction technology based on passive measurement has been widely used in many fields such as the exploration of submarine cultural relics and the detection of submarine pipelines due to its advantages of low cost,simple structure and not easy to be affected by the environment.Because the underwater environment is more complex and unknown than the terrestrial environment,how to better apply the monocular vision system in the passive measurement method to the underwater 3D reconstruction technology has important academic significance and practical value.The purpose of this paper is to apply the monocular vision system to underwater to perform3 D reconstruction of underwater target objects.Finally,a 3D reconstruction system for underwater targets based on monocular vision was successfully built,and the underwater image preprocessing and feature matching in the 3D reconstruction were studied in depth.Specific research contents include:Firstly,the components of the underwater monocular stereo vision acquisition system are expounded,the transformation relationship between the visual coordinate systems is analyzed,and the monocular camera imaging model is constructed.Secondly,the principle of epipolar geometry and the solution process of the fundamental matrix are expounded,which lays a solid theoretical foundation for the subsequent solution of 3D reconstruction.Secondly,a weighted fusion-based underwater image enhancement method is proposed to solve the problems of insufficient clarity of underwater images,blurred edges and lack of details.Divide the input image into three components,and perform the improved adaptive gamma correction for the luminance channel in LAB space to form input component one,and perform histogram equalization and bilateral filtering operations in RGB space to form input component two,in LAB space.The improved Retinex image enhancement operation on the luminance channel forms the input component three,and finally the three components are weighted and fused according to a certain proportion to form the final enhanced image.Then,in view of the problem that the color distortion and low contrast of underwater images reduce the number of feature detections and the accuracy of feature matching,a SIFT algorithm based on Hellinger distance fusion Color information(SIFT-HC)method is proposed..First,a bimodal distribution function is used to minimize the grayscale difference between pixels and the color contrast between pixels to obtain the grayscale image to be matched.Then on this basis,Hellinger distance is used as the similarity measure,and the cosine similarity matching method is used for feature matching,and the effectiveness of the algorithm is verified by experiments.Finally,sparse reconstruction,dense reconstruction,and Poisson surface reconstruction are performed on the matching point pairs obtained by feature detection and matching in order to form the final three-dimensional reconstruction model of underwater targets.And compared with the monocular 3D reconstruction method suitable for land scenes based on the SIFT algorithm and the underwater monocular 3D reconstruction method based on(Oriented fast and rotated brief,ORB)features,the effectiveness of this method is verified.. |
