Ship Detection Algorithms Based On Local Point-plane Contrast Product And Feature Difference Coupling In SAR Imagery

Synthetic Aperture Radar(SAR)has been widely used in many fields,such as target detection,ground object classification,etc,because of its all-day and all-weather imaging capability.Ship detection,as an important branch of target detection,is of great significance to the maintenance of ship navigation safety and maritime rights.At present,ship detection based on SAR images focuses on two directions: marine ship detection and inshore ship detection.For marine ship detection,due to the polarization mode,incident angle,sea breeze and other factors of SAR images,marine ship detection under complex sea conditions becomes a challenging problem.For the detection of inshore ships,the port buildings are highly similar to the inshore ships in terms of gray scale and texture characteristics,which makes the detection of inshore ships a challenging problem.In order to solve the above problems,this thesis studies the detection of marine ships and inshore ships respectively.The main work of this thesis is as follows:1)A marine ship detection algorithm based on local point-plane contrast product is presented.In order to reduce the influence of speckle,this thesis uses the mean shift filtering method to smooth the image.After filtering,the edge between the ship target and the background is relatively blurred under complex sea conditions.Therefore,a local difference enhancement factor is constructed to enhance the edge between the ship and the sea surface.However,the scattering coefficients of ship and sea surface are still similar.In order to further enhance the contrast between ship and sea surface,a local point-plane contrast product algorithm is proposed.In order to solve the adaptive problem of ship detection,an adaptive ratio criterion is proposed to get the ship detection results.The experimental data are based on the SAR images of the Bohai Sea,the Yellow Sea,the East China Sea and the South China Sea.The results show that the proposed algorithm is better than the existing algorithms in quality factor at least 0.04 and running time at least 37%.Experimental results show that the proposed algorithm can effectively detect ship targets under complex sea conditions and is superior to the contrast algorithm in detection performance.2)A inshore ship detection algorithm based on local feature difference coupling is presented.Firstly,Markov random field algorithm is used to segment land and sea,and the segmented image is morphologically processed.After sea-land segmentation,the suspected target area was determined by morphological etching operation.Then,in order to detect suspected ship targets,a local feature difference factor based on superpixel was constructed to extract ship targets.However,some ship targets were misjudged because they were vertically connected with the wharf.In order to solve this problem,this thesis presents a vertical segmentation algorithm between ship and wharf.For the segmented image,a inshore ship detection criterion is given to remove false targets,thus obtaining the ship detection results.The experimental data are based on the SAR images of the coastal areas of Bohai Sea,Yellow Sea,East China Sea and South China Sea.The results show that the proposed algorithm is superior to the contrast algorithm in detection probability at least 0.08,and superior to the contrast algorithm in quality factor at least 0.06.The experimental results show that the proposed algorithm can effectively detect ship targets which are highly similar to inshore buildings in gray and texture features,and can effectively detect ship targets vertically connected to the wharf.The detection performance of this algorithm in inshore ship detection is better than that of the contrast algorithm.