Research On Underwater Positioning Method Based On Stereo Vision

It is necessary to collect the information of underwater related targets and complicated environment when underwater robots are engaged in underwater operations.The primary task is to obtain the precise location information for the operating target to ensure the accuracy of robot operation.The underwater detection equipment with binocular stereo vision is more effective than the traditional sonar and other detection equipment when robots work at close range.So in this paper,we studied the underwater target positioning for complex environment.The article mainly includes the following sections:Firstly,we presented a calibration method of the underwater binocular vision,provides the hard and software environment to build the binocular vision system,and established a binocular vision model.we calculated the internal and external parameters of the binocular cameras by using Zhang's calibration method to carry on the underwater calibration of the binocular cameras.Secondly,we preprocessed underwater images and compared the filtering effect of images which were processed by different methods.after achieving image segment,we extracted the ORB feature points of images and carried on stereo matching base on feature points.We calculated the position relation between the target object and the the binocular cameras by using the coordinate of the feature points.Then,we researched the SLAM algorithm with binocular vision.This section mainly includes three parts: the research of the SLAM algorithm for the mobile robot,the research of the SLAM algorithm for the model of underwater motion and the research of ORB-SLAM2 algorithm based on binocular vision.Finally,we verified the performance of the method of underwater positioning by experiment and analysed the result of the experiment.