| As the issue of land resources depletion and ocean resources competition becomes more and more intense,the ocean will become an important base for human beings.In a wide variety of underwater activities,unmanned underwater vehicles(UUVs)play a great role.Because of the covert needs of the underwater recovery station,UUV needs to return recovery platform independently.Compared to other sensors,visual sensor has the highest accuracy in short range location.So vision is used to tracking guide lights in a short range recovery.This paper takes UUV underwater recovery as the background and aims to study a kind of stable and fast monocular visual tracking and positioning method to meet the close UUV recovery of high precision requirement of docking.According to the characteristics of low light environment under water,light is used as a guide for visual tracking.This paper adopts linear guided lights system which contains reference lights and ordinary lights as guidance to guide UUV.We can get position information containing sway,surge,heading and height according to the guided lights’ position and the distance between each other.One of the most important of the UUV approaching the guided lights is to extract the guided lights accurately.And then we can get the eigenvalues to calculate the location information.This paper studies a method based edge detection to extract image.The method combines the improved morphological gradient filter operator with the cubic spline interpolation algorithm to improve the accuracy of edge.The edge detection algorithm can effectively detect the edge of guided lights and the capsule frame of background.After a series processing of images,we will conduct real-time tracking of guided lights.This paper studies a method of tracking target based on image.The algorithm is the improvement of traditional mean shift algorithm.It weights on the target and background respectively,and can effectively improve the problem of target background interference and target occlusion.However,it also can meet the requirements of real-time docking process.Finally,simulation experiments are implemented in usage of the 11-DOF platform,a monocular camera and the lights guidance system.Line motion and sine motion recovery procedures are designed according to actual tasks. |
