Research On Moving Object Tracking For UUV Recovery Based On Vision

In the development and detection of marine resources,Underwater Unmanned Vehicle(UUV)has been widely used in marine exploration because of their long battery life and strong autonomous ability.The rapid and reliable recovery of UUV is an important process,but the surface recovery of UUV caused by the bad weather conditions in the ocean has had great difficulties,so the recovery of UUV has got more and more attention,the recovery of UUV in the military aspects has a hidden,high speed at recovery and the great military value.When the UUV is recovered at close range,compared with the sonar technology,the high searching precision and the anti-interference ability of optics has become a research hotspot.In this paper,the research of UUV process is carried out,in the process of tracking the moving target with light vision that is in the movement of the mother ship above the recovery dock cabin,the underwater noise,the blocked target light source and so on is researched,main points are as following:Firstly,the devices of the single-vision system used in the process of UUV recovery are described and the calculation of UUV conversion in each coordinate system is discussed.The target light source array composed of three triangular submerged illuminators is discussed and UUV relative to the target light source array relative position and attitude of six degrees of freedom information is calculated.Secondly,for the noise in the water,we use the bilateral filter to filter out the noise in the water and reduce the interference to the target light source.The Sobel operator is used to detect the edge of the target light source precisely.The DOTSU is computationally intensive and time-consuming,while the underwater dynamic target tracking requires high real-time performance,so this paper presents the method combined OTSU and Simulated Annealing Algorithm;the collected template and the latest acquisition and the image of the pre-processing has been completed matching by the use of shape-based template matching algorithm,and ultimately determine the return to the target light source array of three triangular light source in the image position coordinates.Once more,as the UUV in the underwater dynamic tracking of the target light source,the target light source is easily blocked,the space-time context tracking algorithm can not achieve accurate tracking,and even lose the target light source.In this paper,temporal-spatial context is merged with Unscented particle filter.When the target light source is blocked,the image of the target light source is not occluded by the Unscented particle filter,and then the position of the target light source is predicted until the position of the target light source is found and determined,and the position information is transmitted,the temporal and spatial context algorithm is used to keep track.Finally,in order to verify the effectiveness of the UUV tracking algorithm,a simulation experiment was carried out in the pool,the moving targets were observed under the conditions of interference light source,occlusion and so on through the use of small gantry crane,iron platform and nylon rope.Finally,the experimental results show that the proposed algorithm is practical.