Research On Underwater Terrain And Topography Matching Method Based On Sonar Image Processing

Terrain and topography matching(UTTM)is a positioning method to obtain the position information of the moving platform by terrain elevation matching or topography scene matching.It is of great significance to the navigation mission of vehicles.However,due to the variability of the underwater vehicle movement,the insignificant elevation of underwater terrain and the low stability of underwater topography measurements,there is a large gap between the UTTM in underwater and on-land tasks in terms of scope of application and the amount of available methods.This has been a limit to the further improvement of the accuracy of underwater navigation system.This paper aims to meet the demand of underwater passive location technology for accumulated error correction used in autonomous underwater vehicles(AUVs)in long-distance missions.Multi-beam bathymetric sonar and side-scan sonar are used as the measuring equipment for underwater terrain and topography data.Intensive study of how to apply sonar image processing techniques in the feature extraction of underwater terrain and topography is conducted.Based on this,terrain matching is feasible in relative flat underwater areas and the corresponding navigation cells can be identified.Besides,methods of underwater topography matching and rotation degree estimation in local area have also been investigate.This research significantly extends the application of UTTM in underwater environment,and provides a new method for accumulated error correction in underwater navigation.The main work and achievements of this paper includes:1.The principle,method and influence factors of UTTM are investigated in the framework of underwater navigation system.Functionally,underwater navigation system equipped with UTTM is divided into four parts,namely fundamental navigation unit,terrain and topography measuring unit,terrain matching unit and topography matching unit.Based on this,the principle of using UTTM in underwater navigation is summarized.Similarities and differences between UTTM applied on land and in the water are analyzed.Furthermore,the advantages of using multi-beam bathymetric sonar and side-scan sonar in UTTM are explained.Technical framework of UTTM is proposed and possible influence factors are discussed.2.Sonar image processing based underwater terrain matching methods for flat underwater terrain area are investigated.Using multi-beam bathymetric sonar as the detecting equipment,according to correlation theory,underwater terrain matching methods based onimage features are proposed for areas whose elevation is small and varies fast.For seafloor of small depth variance,image texture features are used and a characteristic vector is built for the matching process.For seafloor whose depth varies frequently,edge corner features are chosen to distinguish the terrain images.Actual bathymetric data is used in the positioning tests.The results show that the methods' robustness of noise,rotation and scale are good.The terrain matching methods are feasible for underwater passive positioning.3.Navigability analysis methods of underwater terrain using image features are investigated.Conventional terrain statistics are not suited for 2D terrain images and the information they contains cannot fully reflect the navigability of the underwater area.New image features are designed to obtain the uniqueness of a terrain area.With the combination of fuzzy decision theory,a navigation region selecting method is proposed.Another approach for matching area selection is used which gradually refines the suitable regions by judging their depth features,image information amount and self-similarity.Verification tests are conducted with actual bathymetric data and the proposed terrain matching methods.The results show that in the selected navigation cells,positioning error of the matching methods are lower than other zones,which proves the effectiveness of the navigability analysis methods.4.Underwater topography matching and course estimation method using image processing is investigated.Resolution of reference image and real-time image are fixed to reduce the impact of different view point.Circular template and Hu invariant moments are combined to overcome the effect of relative rotation between the matching image pairs.Log-polar transformation is executed to estimate the rotation so that the course information is available.Verification tests are conducted with actual side-scan images.The results show that the matching method is feasible in underwater areas of abundant landform or including obvious marks.This provides a new thought for the application side-scan sonar in underwater topography matching.5.With the combination of the proposed methods mentioned above,an application strategy of UTTM using multi-beam bathymetric sonar and side-scan sonar is proposed.Actual bathymetric data and side-scan image are used to verify the feasibility of applying UTTM in the long-distance underwater missions and provide a frame of the possible research directions.This paper intensively investigated the principle of UTTM and the implementation methods of using sonar image processing techniques in it.Study results show that the proposed methods can promise good stability and are adaptable to various terrain andtopography areas.They are able to be used in the navigation of long-distance underwater missions.The techniques are proved to have high practicability.