On Visual Guidance And Positioning For UUV Recovery

The Nineteenth National Congress of the Communist Party of China put forward the strategic plan of "speeding up the construction of a powerful ocean country." As a marine capital development and security safeguard Important means,UUV is the realization of ocean strength Important weapons and equipment.Among them,the autonomous recovery task of UUV is the key technology of UUV research,and the underwater environment is complex changeable,which brings great difficulties to the recovery of UUV.considering the high accuracy of visual positioning at close range and its successful application on land,this paper studies the visual guided positioning of UUV recovery process in the background of fork-type UUV recovery.The main research contents are as follows:Firstly,based on the scattering model,a single channel underwater turbidity image enhancement method without prior information is proposed.The underwater image is modeled and analyzed;According to the law of imaging and the analysis of experiments,this paper proposes a background light estimation method based on local block entropy analysis;The optimal transmittance is designed by weighing clarity and image fidelity cost function through improved weighted least squares edge-preserving filtering algorithm.On this basis,the adaptive adjustment factor is designed to modify the image enhancement formula.finally,the effectiveness of the proposed algorithm is verified by comparing the results of multi-scene experiments,which can effectively enhance the UUV visual guidance localization image.Secondly,the UUV visual guidance positioning markers are selected and verified,and the improved OTSU algorithm is designed to solve the target extraction problem of UUV visual guidance positioning under different background light conditions.Based on the UUV forkcolumn recovery,an asymmetric L-shaped light array guide marker for detection and positioning was designed;An OTSU method for self-detection of threshold interval is designed for threshold segmentation of foreground image under variable background light;The pseudolight source is eliminated by combining support vector machine(SVM)with Blob morphological operator;The experiment proves the feasibility of the detection algorithm under different background light conditions.Modeling the UUV vision guidance system and completing the six-degree-of-freedom positioning method of UUV monocular vision under the condition of no occlusion;The problems of single light source occlusion(25% target occlusion)and two light sources(50% target occlusion)were discussed respectively.The visual positioning method of UUV which can deal with the occlusion of two light sources is designed called tracking after single target segmentation method;Multi-objective tracking method using Hungarian matching(hm)combined with Kalman filter is used to verify the proposed method the validity of the law.Finally,the UUV visual guided positioning recovery experiment was designed.The environment and equipment of the experiment are introduced in detail;the strategy flow of designing UUV visual guidance positioning execution is designed.Pool experiment and data analysis verify that the proposed UUV visual guided positioning method can be competent for a certain degree of background light variation and up to 50% guided target occlusion of the UUV visual recovery task.