SAR Ship Detection Based On CNN And Its Application In Mobile Terminal

For the shipwreck accident,the staff of the Marine Search and Rescue Center uses the Synthetic Aperture Radar(SAR)image to search the target of ship.In order to ensure the accuracy of the Ship positioning,they usually use the way of using the manual annotation for ship target,which not only is time-consuming,but also can't meet the needs of obtaining the real-time search and rescue information.This usually leads to the staff of the Marine Search and Rescue Center missing the best rescue time.At the same time,the Chinese Academy of Sciences Institute of Remote Sensing and Digital Earth Institute of high score major issues raised the Interoperable working mode of satellite radar,emergency center and mobile terminal,which can help Marine Search and Rescue Center easier to obtain accurate rescue information.In this paper,the detection algorithm of Synthetic Aperture Radar(SAR)image ship target and its application in mobile terminal are explored under the background of the application of the demonstration system of Earth's surface system scientific research application system.The main contributions of this paper are as follows:(1)Aiming at the requirement of ship detection for complex SAR images,this paper designs a SAR image detection algorithm based on the combination of significance detection and Convolution Neural Network(CNN).By collecting SAR images of ship targets established CNN training dataset,land and sea mask significant regional extraction,CNN ship detection step for the detection of high-resolution SAR images.The detection algorithm of the ship detection rate of up to 95%,missed the test rate as low as5%,false alarm rate as low as 9%,quality factor of 0.91,to meet the basic needs of the target search.(2)In order to solve the problem that the mobile terminal is loading the five-layer fifteen tiles,the fast loading algorithm of SAR image is designed for mobile terminal.The algorithm displays the SAR image tiles which after ship detection on the mobile device by means of on-demand dynamic loading,cyclic loading display,caching and delayed deletion.The experimental results show that the average loading time of the tile is reduced by 47%,and the stability of the memory consumption of the mobile device is guaranteed.(3)This paper realizes the related functions of SAR image ship detection for mobile terminal,and strengthens the connection between the mobile end system and the product production subsystem of the application system of Earth surface system scientific research and application.Making the satellite radar,emergency center,mobile terminal three work mode of mutual exchange become a reality.In the application system of the Earth's surface system scientific research application,the detection algorithm of the ship detection module in this paper satisfies the application level of the SAR image,and the display module of the mobile end subsystem satisfies the actual demand of the user to browse the test data.The study of this paper provides a great help for the search and rescue of the ship.