Ship Detection And Motion Monitoring With Geosynchronous Satellite Remote Sensing Images

Ocean,as an important part of the human living environment,has always been occupying an extremely important strategic position both in the national economy,development,and in the implementation of China's Matitime Silk Road Strategy.Over the sea surface,ship detection and monitoring is an essential task,not only for the development of the economic,the protection of marine environment,but also for the management of the marine area,as well as the maintenance of the sea rights and interests.With the increasing intensity of marine development,as well as the increasingly serious situation of maritime ownership disputes,sea surface ship detection and dynamic monitoring applications put forward higher requirements on the real-time and mobility of the monitor and control system on satellite.Satellite remote sensing has been proved to be an effective and efficient technique for ship detection,classification,recognition,and motion feature extraction,yet for low-orbit satellites,it is now insufficient for large-scale,high-frequency and real-time dynamic monitoring.Comparatively,from geosynchronous-orbit,high-resolution remote sensing could continuously observe the marine surface and acquire large-scale ship motion information,hence great potential lies in this kind of satellites for ship detection and monitoring application.Up to now,high-resolution geosynchronous remote sensing is still at the development stage,and hasn't gone further study into the application of sea surface ship detection and monitoring.In order to achieve near real-time observations as soon as possible,this thesis thoroughly investigated the spatial-temporal imaging characteristics and application potentials of the optical geosynchronous-orbit satellite,GF-4.On the basis of this,a gradient threshold based ship detection method was established and successfully performed on a single-band image of GF-4.Combined with this ship detection method and the band-to-band time lags of the GF-4 multi-spectral imagery,we proposed a ship motion detection method by using only a single GF-4 multi-spectral image.Furthermore,the traditional ship motion monitoring method based on image sequences was improved with the introduction of the ship motion detection method we proposed.Since time and weather restrictions could not be ignored while imaging with active optical remote sensing sensors,for all-weather and all-time observations of the sea surface ships,this paper further studied the imaging characteristics and application potentials of the geosynchronous-orbit synthetic aperture radar satellite.On this basis,time-domain geosynchronous SAR imaging simulation was implemented for the scenes where exists ships with some known motion features.The ships were then successfully identified from the simulated images with a well-known CFAR ship detection method.In view that the sea clutter during such a long synthetic aperture time of geosynchronous SAR may severely impact the imaging quality and subsequent ship detection,time-domain SAR imaging simulation were also carried out on both the timevarying and the frozen rough sea surface.The amplitude frequencies of the simulated geosynchronous SAR images for these two kinds of sea surface were then tallied,and the results of comparisons showed their similarity in statistics,which was a valuable conclusion for future geosynchronous SAR system design and sea surface ship detection.The main research aspects included in this dissertation are as follows:1.According to the time-share observing mechanism of GF-4 satellite,the bandto-band imaging interval of multi-spectral imagery was calculated by using the image displacement of moving ships and their velocities from AIS records.2.In accordance with the image features of ships in GF-4 multi-spectral imagery,a gradient threshold based single-band ship detection method was established,and the position,velocity and direction accuracies obtained were evaluated by using corresponding AIS data.3.Based on the single-band ship detection method,and by multi-band collaborative detection,a single GF-4 multi-spectral dataset based ship motion feature extraction method was proposed,and the velocities and directions gained by this method were also evaluated by using corresponding AIS data.4.On the basis of the single multi-spectral imagery ship motion detection method,and taking the advantage of the staring imaging mode of GF-4,a new time sequential imagery method for ship dynamic monitoring was presented,and the velocities and directions achieved by this method were evaluated by the traditional time sequential imagery monitoring method.5.Orbit simulation and numerical calculation were conducted for the investigations on the observation characteristics of geosynchronous SAR and its potential in ship detection.Then geosynchronous SAR time-domain simulations of moving ships were carried out.The resulting images were analyzed,and the ships were successfully detected by the method of CFAR.6.The time-varying electromagnetic scattering scenes of rough sea surface were built up and then used for geosynchronous SAR imaging simulation.The results of simulation and their impacts on ship detections were also analyzed.The innovative achievements this research made are:1.The single GF-4 multi-spectral imagery ship motion detection method was proposed for the first time.And by this method,high-precision and high-efficiency ship motion detection was performed in this study.2.The geosynchronous SAR imaging simulation of moving ships were carried out,which to a great extent proved that this kind of SAR,one with the characteristics of long synthetic aperture time and curved observation orbit,could be utilized for moving targets detection.3.The characteristics of the geosynchronous SAR images and their amplitude statistics were studied,and this study would provide first-hand information both for future design of a sea surface ship detection geosynchronous SAR system,and for the development of corresponding imaging and ship detection methods.