Research On Sea-sky-line And Small Ship Target Detection Method Based On Unmanned Surface Vessel

In recent years,sea-sky-line and small ship target detection technology based on visible image of unmanned surface vessel(USV)has gradually become a research hotspot in the field of USV.Compared with radar and infrared images,visible images have the advantages of low cost and high resolution.It is difficult to detect the ship targets because of the lack of feature information such as size,shape and texture,and easy to be disturbed by sea clutter and clouds.In this paper,the visible image of sea-sky background is analyzed experimentally with USV as the carrier,and the area of sea-sky-line is segmented by locating sea-sky-line,and then achieve the detection of ship target.Firstly,several common image preprocessing methods are introduced,including image gray transform,geometric transform,image smoothing,edge detection and morphological filtering;and then the composition of USV vision system is discussed.In view of the low visibility scenarios such as sea fog and rainfall,several image defogging methods are studied,and an adaptive MSR algorithm is proposed,which effectively improves the image signal-to-noise ratio and color distortion.A sea-sky-line detection algorithm based on the adaptive MSR algorithm and Hough transformation is proposed.By thinning the edge and reducing the angle range of theta,the problem of large Hough transformation calculation is solved.Experiments show that the success rate of sea-sky-line detection is about 95%.In this paper,the rotation correction is achieved by tilt angle of sea-sky-line,and according to the position of the rotating sea-sky-line,thirty pixels height above the sea-sky-line and twenty pixels height below the sea-sky-line are segmented into the sea-sky-line region.Finally,Local OTSU algorithm based on background suppression and GrabCut algorithm are used to detect ship targets.Simulation experiments show that the local OTSU algorithm based on background suppression can effectively enhance the target feature information and suppress the interference of background noise,and the success rate of target detection is about 94%.The local OTSU algorithm based on background suppression has a good application prospect in the field of target detection of USV.