Research On Underwater AUV Navigation And Target Detection Based On Acousto-optic Image

When autonomous underwater vehicle completes underwater task,it mostly uses inertial measurement unit or inertial navigation system to assist positioning.The inertial navigation system is based on reckoning.The attitude estimation of any position system based on row reckoning will accumulate with the increase of time,which will lead to the continuous drift of position data and affect the task completion of AUV.Ocean sonar imaging is decades earlier than AUV,it is a very powerful technology.Underwater acoustic imaging is a part of the field of acoustic imaging in a wider range,which can be used to provide underwater observation and inspection functions when the optical system is unable to achieve.In this paper,based on the underwater acoustic image target detection,after preprocessing the collected underwater acoustic image,the computer vision saliency detection algorithm is improved and applied to the underwater acoustic image,the underwater acoustic image target saliency detection is studied,and the dynamic and static state of the underwater acoustic image after the target saliency detection is judged,As a new anchor node for AUV,it can be used as a reference point to reduce the accumulated error of AUV inertial navigation.After detecting a static target as a reference point,the AUV can obtain multiple distance information from the sonar image.The change of distance can reflect the relative motion of the AUV relative to the reference point.In this process,based on the calculation of the extended Kalman filter,the accumulated error of the AUV inertial navigation can be reduced.Due to the complex and changeable characteristics of the underwater medium and its boundary in the underwater acoustic channel,and the projection characteristics of the acoustic wave itself,the underwater acoustic image collected often has the characteristics of strong noise,serious distortion,fuzzy edge of the target,low resolution,even if the image of the underwater acoustic target is preprocessed,the color,texture and other information of the target can not be obtained accurately.After image enhancement preprocessing,the color,texture,shape and other information of underwater optical image are relatively accurate.However,the underwater location based on visual image is more complex and less real-time than that based on acoustic image.The results of underwater acoustic image and optical image matching can enrich the distance information of optical image target,so as to assist the visual inertial navigation;it can also be applied to match the target when sonar equipment and optical equipment switch under the influence of visual distance and ocean noise to ensure continuous target tracking.